Towards Supporting Security and Privacy for Social IoT...
Transcript of Towards Supporting Security and Privacy for Social IoT...
Research ArticleTowards Supporting Security and Privacy for Social IoTApplications A Network Virtualization Perspective
Jian Sun 1 Guanhua Huang1 Arun Kumar Sangaiah2
Guangyang Zhu1 and Xiaojiang Du3
1Key Lab of Optical Fiber Sensing and Communications (Ministry of Education) UESTC Chengdu China2Vellore Institute of Technology Tamil Nadu India3Department of Computer and Information Sciences Temple University Philadelphia PA USA
Correspondence should be addressed to Jian Sun sjuestceducn
Received 16 January 2019 Accepted 25 February 2019 Published 14 March 2019
Guest Editor Houbing Song
Copyright copy 2019 Jian Sun et alThis is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
Network function virtualization (NFV) is a new way to provide services to users in a network Different from dedicated hardwarethat realizes the network functions for an IoT application the network function of an NFV network is executed on general serversand in order to achieve complete network functions service function chaining (SFC) chains virtual network functions to worktogether to support an IoT application In this paper we focus on amain challenge in this domain ie resource efficient provisioningfor social IoT application oriented SFC requests We propose an online SFC deployment algorithm based on the layered strategiesof physical networks and an evaluation of physical network nodes which can efficiently reduce bandwidth resource consumption(OSFCD-LSEM) and support the security and privacy of social IoT applications The results of our simulation show that ourproposed algorithm improves the bandwidth carrying rate time efficiency and acceptance rate by 50 60 and 15 respectively
1 Introduction
In traditional service provider networks network functions(NFs) (eg intrusion detection systems (IDSs) gatewaysload balancers network address translators (NATs) and fire-walls [1 2]) for social IoT applications [3ndash6] are implementedby specialized hardware devices and it is expensive to incor-porate some new devices into an existing service networkWith an increasing number of social IoT applications thedemand of corresponding NFs for each user necessitatesa significant amount of hardware resources Furthermoresecurity and privacy preservation in social IoT applicationsare also important issues that need to be solved [7ndash10]To address these problems network function virtualization(NFV) technology has been put forward In NFV networksNFs are implemented in the form of software instead ofdedicated hardware and separately run on different virtualmachines (VMs) [11 12] and thus guarantee the security andprivacy requirements As shown in Figure 1 NFs that run inNFV networks are called virtual network functions (VNFs)
Different VNFs are arranged in a prescribed order to forma service function chain (SFC) to fulfill the communicationrequirements [13ndash15]
NFV is convenient for social IoT service providers todeploy VNFs on commercial servers and manage the under-lying network [16ndash18] Using NFV technology the socialIoT service provider can flexibly deploy virtual NFs oncommercial servers [19ndash21] which are traditionally fulfilledon dedicated hardware Furthermore with an increasingsocial IoT service demand [22ndash25] common commercialservers can provide support to multiple VNFs which sig-nificantly reduces the vacancy rate of physical resources[26ndash28] and saves the cost of purchasing new dedicatedhardware The network operators can flexibly deploy andchain the VNFs according to the social IoT usersrsquo requestsand practical network topology This approach decreasesthe use of inapplicable functions and the correspondingdeployment costs Moreover this approach can customize thesocial IoT services for the clients introduce a broad busi-ness outlook and promote its commercialization Regarding
HindawiSecurity and Communication NetworksVolume 2019 Article ID 4074272 15 pageshttpsdoiorg10115520194074272
2 Security and Communication Networks
Physical Resources
Virtualized Resource Pool
VNF Deployment and Chaining
VNF1-NAT
VNF2-IDS
VNF4-Firewall
VNF3-Load
Balance
SFC
User User
NAT IDS FirewallLoad Balance
Server Server
Server Server
VM
VM
Server Server
Server Server
VM
VMVM
VM
VM
VM
Figure 1 SFC organization by virtualizing commercial servers
traditional NFs VNFs increase policy compliance capabil-ities [29] and promote the emergence of more novel tech-niques and the development of NFV for social IoT applica-tion
NFV can bring additional benefits to social IoT ser-vice providers ie optimize their operational expenditure(OPEX) and capital expenditure (CAPEX) [11 30ndash32] andimprove the quality of service (QoS) by decreasing latencyreducing bandwidth consumption and increasing adaptationNFV introduces many benefits to both users and social IoTservice providers but there are still many research limitationsthat need to be dealt with For example the bandwidthconsumption of the SFC is an important research topic
Bandwidth consumption represents the resource cost ofcommunication among usersclients [33 34] The networkoperator seeks to achieve low bandwidth consumption tocomplete the deployment of the current SFC request and saveresources for the next SFC request to improve the accommo-dating capacity of the network [35]With an increasing diver-sity of service requirements and additional pressure frommassive information transmission bandwidth resources havebecome progressively scarce [36] Furthermore low band-width consumption can provide the clientuser a goodexperience due to the excellent performance of the networkoperator In previous research several algorithms have aimedto reduce bandwidth consumption In [17] the authors
Security and Communication Networks 3
focused on bandwidth consumption optimization and pro-vided a framework for their studied problem Thus insteadof the existing hardware environment targeted researchof NFV can produce increased benefits and reduce theenergy and space consumption of various middle boxes[19 37]
In this paper we study the problem of how to provisionsocial IoT oriented SFC requests while minimizing the band-width resource consumption as well supporting the securityand privacy requirements The studied problem has beenproven to be an NP-hard problem [35 37 38]
Thus we propose an efficient algorithm called OSFCD-LSEM with layered strategies for physical networks It notonly efficiently solves the SFC deployment problem but alsomakes the physical networkmore robust Our OSFCD-LSEMalgorithm evaluates the nodes in the physical network tooptimize the bandwidth consumption and thus achievesa higher acceptance ratio and shorter response time (ielatency) for user demands than the exiting work In additionwe can make the physical network more robust using ourOSFCD-LSEM algorithm The main contributions of thispaper are as follows
(i) We develop a model to evaluate the physical networkand extend it precisely to its ldquoweak pointsrdquo and makethe physical network more robust
(ii) We propose an efficient algorithm (ie OSFCD-LSEM) for social IoT oriented SFC deployment that isbased on the layered strategies of a physical networkand evaluation of a physical network node to optimizethe consumption of bandwidth resources and guaran-tee the security and privacy
(iii) We conduct extensive simulations to verify and eval-uate our algorithm in this paper The results showthat our proposed algorithm can efficiently addressthe online SFC deployment problem
The remainder of this paper is organized as followsSection 2 provides an overview of related works Section 3provides the problem descriptions and formulation InSection 4 we describe our SFC deployment algorithm basedon the layered strategies of physical networks and eval-uation of physical network nodes Section 5 shows thesimulation results and analysis Section 6 summarizes thispaper
2 Related Work
NFV aims to satisfy client demands with minimal resourceconsumptions (eg bandwidth consumption and computingcore consumption) and high performance (eg low latencyand high throughput) [20] An increasing number of studieshave focused on the more efficient deployment of social IoToriented SFC requests in various scenarios
The research in [39] focused on the suitability of differentarchitectures of data center for a resilient SFC and theplacement of VNFs with high availability constraints In[40] the joint VNF placement and path selection problemwere studied The authors proposed a chain deployment
algorithm to balance the chain length and reuse factor itstarget was to serve as many demands as possible underlimited resources The research in [41] designed a heuristicalgorithm to address the NFV-RA problem in a coordinatedmanner by splitting the SFC request In [42] the authormade a BCMP mixed queuing network to replace the SFCmodel and proposed the convex optimization problem toshorten the acceptance interval of the service chain Thesimulation results shown the algorithm in [42] can effec-tively adapt the resource usage to the network dynamicsand serve more demands than other existing algorithmsThe authors of [43] studied SFC deployment in a multi-domain network and proposed a vertex-centric distributedcomputing algorithm to find all feasible SFC deploymentmethods of user requests in the distributed orchestrationframework then the algorithm selects the most appropriatescheme to deploy the SFC to achieve efficient performanceIn [44] the researcher discussed the phenomenon that theVNFs may sprawl across the network due to inefficientmapping during the SFC orchestration process and designedan efficient greedy heuristic algorithm to solve the problemThe author in [45] studied the service availability constraintsin a data center and designed a heuristic algorithm to deploythe SFC In the situation of distributed NFVs the authors in[46] studied the security level of a network security frame-work The authors proposed a new optimization algorithmto avoid a single point of failure in a bottleneck problemand obtained reasonable performance relative to that of acommon device Li et al [47] presented a real-time resource-distributing system for NFV which integrates timing analysiswith other algorithms such as timing abstraction SFC con-solidation and linear programming algorithm to efficientlydistribute network resources while considering latency con-straints
Many researchers have studied the deployment of thesocial IoT oriented VNF or SFC and a few of them haveattached importance to the total bandwidth consumptionYe et al [17] focused on the joint topology design and SFCdeployment problem to minimize the total consumption ofbandwidth resources Their main idea was to reduce band-width usage and cost by prioritizing VNF deployment In[48] the authors proposed and analyzed two deploymentstrategies HSD and VSD whose performance was analyzedin terms of cost They claimed that HSD had better per-formance in terms of an even distribution of the load overall servers access links and ToR switches The authors in[49] combined NFV with cloud computing and designeda bandwidth-guaranteed SFC-placing algorithm then theauthors tested the performance of their algorithm in a datacenter The simulation results showed that the heuristicalgorithm had excellent performance In [38] the authorsdesigned a forecast-assisted online SFC deployment algo-rithm that included the prediction of future VNF require-ments The simulation results showed that the algorithm in[38] could reduce the blocking probability of SFC requestsand effectively improve the profit of the service provider fromSFC deployment
Overall bandwidth is the most prominently expensiveresource that directly affects the number of demands that the
4 Security and Communication Networks
Client Client
Client
Client
90
f150
f270
f380 90
N4N5N3
N1
N2 N6
N7
N8
70
7080
8050
(a)
Client
Client
8070
70
N4N5N3
N1
N2 N6
N7
N8
Client Clientf150
f270
f380 90
(b)
Figure 2 Two examples of SFC deployment
network can satisfy online and the capacity of the physicalnetworkThus in this research we focused on the bandwidthconsumption and designed an efficient algorithm to provisionsocial IoT oriented SFC requests
3 Problem Formulation
In this paper we studied the online social IoT oriented SFCdeployment problem We considered a situation in whicheach SFC request has a pair of start and terminal nodesattached to the given physical network nodes and a specificsequence of VNFs that make up continuous functions Wehave to deploy these VNFs onto the corresponding physicalnodes and then organize the VNFs to form an SFC Toreduce the bandwidth resource consumption we need to usefewer nodes to shorten the length of the SFC as much aspossible
A user request can be denoted as 119878 = (119865S 119864S)where 119865S = 1198911 1198912 119891m is the set of VNFs and119864S = 1198901 1198902 119890q represents the virtual link after SFCdeployment The real physical underlying network can bemodeled as G = (N L) where G is an undirected weightedgraph 119873 = 11987311198732 119873y is the set of physical nodesand 119871 = 1198711 1198712 119871k is the set of real links inthe physical underlying network We use 119875 to denote thephysical path that holds the SFC request We define 119862119879119861 todenote the total bandwidth consumption which is definedin
119862119879119861 = sum119871 119894 119890119895isin119875 119878
119862119871 119894 119890119895119861 (1)
where119871 119894 119890119895 denotes that virtual link 119890119895 is deployed on the linkLi and 119862119871 119894 119890119895119861 is the bandwidth consumption of virtual link 119890119895which is deployed on the linkLi Wedefine119877119873119894119862 as the availablecomputing resources of the physical node119873119894 and 119862119873119894 119891119895119873 as thecomputing resource requirements of VNF 119891119895 which wouldbe deployed on the node Ni 119877119871 119894119861 is the available bandwidthresources of the physical link 119871119894
To deploy an SFC we must map the VNFs to some nodeand the virtual links 119864S to some real links in the physicalnetwork the path 119875 119878 which would hold the SFCs musthave sufficient nodes and computing resources to deploythe corresponding VNFs and the physical links must havesufficient available bandwidth for communication amongthose VNFs In addition the available bandwidth resourcesmust satisfy the requirements of the 119864S in the correspondinguser request Similar to [10 20] this paper assumes thateach VNF from the same SFC needs to use different physicalnetwork nodes Then the number of VNFs (denoted as Fs)should be less than the number of nodes of physical pathP 119878(denoted by 119875 119878) Then the problem of SFC deploymentcan be formulated as follows
min sum119871 119894 119890119895isin119875 119878
119862119871 119894 119890119895119861119904119905 forall119871 119894 119890119895 isin 119875 119878
119877119871 119894119861 minus 119862119871 119894 119890119895119861 ge 0forall119873119894 119891119895 isin 119875 119878119877119873119894119862 minus 119862119873119894 119891119895119873 ge 0119875 119878 minus 119865119904 ge 0
(2)
Formula (2) is used to model the SFC deployment prob-lem which can minimize the bandwidth consumptions whilefinishing SFC deployment There must be sufficient availablecomputing resources to deploy corresponding SFCs and thebandwidth must be sufficient to satisfy the communicationdemands among the corresponding VNFs In addition theremust be sufficient nodes in the physical path P to deploy theVNFs of the SFC request
Figure 2 shows two different examples of provisioningan SFC request both of which successfully deploy the SFCrequest and satisfy the clientsrsquo demands Nodes N1 and N8are the requested client node and destination client noderespectively There are three VNFs (ie f 1 f 2 and f 3) inthe SFC request Client node N1 needs to transmit 50 units
Security and Communication Networks 5
of traffic to VNF f 1 Between VNFs f 1 and f 2 70 unitsof information need to be transmitted Between VNFs f 2and f 3 80 units of traffic need to be transmitted Thereis a 90-unit transmission demand from VNF f 3 to thedestination client node N8 As shown in Figure 2(a) wedeploy the VNFs f 1 f 2 and f 3 onto the physical nodesN2 N5 and N6 respectively Then we find the path P =1198731-11987321198732-11987331198733-11987351198735-11987371198737-11987361198736-1198738 to hostthe SFC In this deployment scheme the total bandwidthconsumption is 440 units Although this approach success-fully deploys the SFC request it wastes bandwidth resourcesIn Figure 2(b) the VNFs f 1 f 2 and f 3 are deployed ontothe physical nodes N1 N6 and N8 respectively Then wefind a short path P = 1198731-11987321198732-11987361198736-1198738 (shown asthe red line in Figure 2(b)) which can also deploy all VNFsfrom the user requests However the total consumptionof bandwidth resources of this scheme is only 220 unitswhich is approximately half of the consumption of thedeployment in Figure 2(a) In Figure 2(b) path P is theshortest path to deploy this SFC request and this schemecan reduce more bandwidth consumption Mapping theSFC to the path in Figure 2(b) the service network candeploy more SFC with other links which is impossible inFigure 2(a)
Therefore different deployment schemes significantlyaffect the bandwidth consumptions and thus affect the avail-able network capacity An efficient algorithm is important andurgently needed to better deploy SFC requests and reducebandwidth consumption
4 Algorithm Design
Since the optimal SFC deployment problem is NP-hard [3537] we design an efficient algorithm for solving our researchproblem in this section Our proposed algorithm employsthe layered strategies of physical networks and evaluationof physical network nodes to optimize the consumption ofbandwidth resources which is denoted by OSFCD-LSEMThe basic idea of OSFCD-LSEM is to find the shortest pathsto deploy SFC requests while saving as many bandwidthresources as possible to satisfy more user requests Whena user demand arrives the OSFCD-LSEM algorithm beginsto handle it First calling Algorithm 2 the OSFCD-LSEMalgorithm layers the physical underlying network and obtainsthe information from the nodes and links in each layer ofthe physical network Then it calls Algorithm 3 to performan evaluation of nodes in the network and select some nodesthat are most suitable to host the VNFs of this SFC requestFinally it connects the deployed VNFs by using a shortestpath to fulfill an SFC By layered strategies and selection ofthe nodes for VNF deployment the OSFCD-LSEMalgorithmcan deploy VNFs in the most suitable nodes and deploythe SFC requests in appropriate simple paths to save morebandwidth resources The physical path must contain thestart node Nr and the terminal node Na Furthermore thenodes in the physical path must have sufficient resources tohost the VNFs of the user request
In the proposed OSFCD-LSEM algorithm we providesome definitions of variables 119866119871 is the physical network after
layering and 119881119883 is used to denote the set of nodes in the X-th layer The X-th layer is denoted as LX 119866119883119871 is the innerlayered network in the X-th layer (LX) and we used LY todenote the Y-th layer in 119866119883119871 119881119894(119883119884) is the set of nodes in theLY of 119866119883119871 which includes the node Ni 119864119883 is the set of linksthat connect the nodes from LX and LX-1 119864119894(119883119884) denotesthe corresponding links that connect the nodes in the LY-1about nodeNi and 119871119894119872119860119883 is the correspondingmaximal layerof node Ni LMAX is the layer number of GL 119873119879 is the nodenumber of the physical network and 119871119879 is the link numberof the physical network G The OSFCD-LSEM algorithm isshown in Algorithm 1
Next Algorithm 2 is responsible for handling hierarchicalphysical networks in our algorithm Algorithm 2 layers theentire network to obtain the layering information of thenodes and links in the network and outputs the results toother algorithms Therefore Algorithm 2 is the basis of ourSFC request deployment scheme and physical network nodeevaluation The physical network layering can be formulatedas follows
119866119871 =119871119872119860119883sum119883=1
(119881119883 119864119883) +119871119872119860119883sum119883=2
119866119883119871 (3)
119866119883119871 = sum119881119894isin119881119883
119871119894119872119860119883sum119884=1
(119881119894(119883119884) 119864119894(119883119884)) (4)
119871119872119860119883sum119883=1
119881119883 minus 119873119879 ge 0 (5)
119871119872119860119883sum119883=2
119864119883 +119871119872119860119883sum119883=2
sum119881119894isin119881119883
119871119894119872119860119883sum119884=2
119864119894(119883119884) minus 119871119879 = 0 (6)
In (3) two parts make up 119866119871 one part is the inner layernetwork 119866119883119871 about the X-th layer (LX) and the other partis the overall layered network The layering process beginsfrom the request node 119873119903 thus 1198811 = 119873119903 1198641 = 0 and1198661119871 = 0 Equation (4) shows that to make 119866119871 closer to thephysical network G each layer except layer L1 should obtainits inner layer node-related information The OSFCD-LSEMalgorithmcanobtain amore precise evaluation of the physicalnetwork and more efficiently deploy the corresponding SFCrequests In (5) the equation describes that all nodes in thelayered physical network must be in the corresponding layerIn (6) the equation shows that each link should be in thecorresponding layer or inner layer
In Figure 3 we give an example of layering a networkFigure 3(a) shows the topology of a physical network andit has a total of 10 nodes Figure 3(b) provides the detailedprocessing procedure for layering the network topology Weassume that the start node 119873119903 in the request is node N1 andthat the terminal node 119873119886 in the request is node N9 Firstour algorithm places the start node N1 into L1 (N1 is theonly node in V1 of L1) and places nodes N2 N3 and N4into L2 because they directly have links to node N1 Thenour algorithm places nodesN5N6 andN9 all of which havelinks to some nodes in the L2 (ie nodes N2 N3 and N4)
6 Security and Communication Networks
Input (1) SFC request (2) physical network 119866Output SFC deployment scheme(1) Receive a user request(2) Initialize Path = [](3) 119873119886 997888rarr Path119873119871 =119873119886(4) Layer the topology Algorithm 2(119873119903119873119871 119866)(5) Get 119871119860 the layers that destination client belongs to(6) while 119871119878 gt max(119871119860) + sum119871119872119860119883119883=1 max(119871119894119872119860119883) forall119873119894 isin 119881119883 do(7) if max119871119860 = 119871119872119860119883(8) 119873119879119864119872119875 = Algorithm 3( true max119871119860)(9) 119873119879119864119872119875 997888rarr Path(10) 119873119871 =119873119879119864119872119875(11) else(12) 119873119879119864119872119875 = Algorithm 3( false max119871119860)(13) 119873119879119864119872119875 997888rarr Path(14) 119873119871=119873119879119864119872119875(15) end if(16) 119871119878 = 119871119878 - 1(17) VNF 997888rarr 119873119879119864119872119875(18) Algorithm 2(119873119903119873119871 119866)(19) Update 119871119860(20) end while(21) if 119871119878 lemax(119871119860)(22) Select min LX isin 119871119860 ampamp LX gt 119871119878(23) while 119873119903 notin Path do(24) 119873119879119864119872119875 = Algorithm 3 ( true LX)(25) 119873119879119864119872119875 997888rarr Path(26) 119873119871 =119873119879119864119872119875(27) LX = LX - 1(28) VNF997888rarr 119873119879119864119872119875(29) end while(30) end if(31) SFC deployment scheme is Path
Algorithm 1 OSFCD-LSEM algorithm
into L3 (we specify that all nodes can only belong to onelayer except the terminal node N9 thus node N4 cannot bepart of L3 even though it connects with node N3 which isin L2)
In our layered network except for the destination clientnode N9 the nodes in one layer must have links to the nodesin the previous layer Thus nodes N7 N8 and N10 directlyhave links to the nodes in L3 (ie nodes N5 N6 and N9)whereas N10 connects only with destination client node N9among the three nodes in L3 NodeN10 should not be placedin layer L4 Thus we place only nodes N7 and N8 into L4Nodes N9 and N10 connect with nodes N7 and N8 in L4thus we place nodes N9 and N10 in layer L5 and becauseN9 connects with node N10 our algorithm places node N9in L6 When ten nodes in G belong to the correspondinglayers the overall network-layered processing finishes Thusthe OSFCD-LSEM algorithm can guarantee finding a pathwithout loops For each LX we also need to layer and obtainthe inner layer 119866119883119871 In the example shown by Figure 3(b) thelayer L2 has an inner layer 1198662119871 that includes two layers For1198662119871 in Figure 3(b) each layer X le LMAX and each node Ni isinVX should be set as the start node 119873119903 let 119873119886 = 0 Then we
obtain their inner layer information In 1198662119871 both 1198711198733119872119860119883 and1198711198734119872119860119883 are equal to 2 whereas 1198711198732119872119860119883 is equal to 1Overall the physical network is layered into six layers
with two inner layer networks associated with nodes N3 andN4The start node119873119903 is the only one in the first layer and theterminal node119873119886 is in three layers including L3 L5 and L6Thus we obtain three paths that can be used between119873119903 and119873119886 119871119875 is used to denote the length of path which is equal tothe number of VNFs that the path can hold LS is the lengthof an SFC which is equal to the VNF number of the SFCrequest
Algorithm 3 evaluates the nodes in the physical networkand selects a node which is most suitable to deploy therequired VNF After obtaining the layering informationAlgorithm 3makes the decision regarding whether themulti-ple links can satisfy the user request When themaximal layernumber in 119871119860 of the terminal nodes 119873119886 which is obtainedfrom the sumof all inner layers is smaller than the SFC length119871119878 in the user request the physical network cannot satisfy theuser request For example if we need to deploy an SFC intothe physical network in Figure 3(a) the start and terminalnodes are N1 and N9 The maximum layer number in 119871119860 is
Security and Communication Networks 7
Input (1) 119873119903 (2) 119873119886 (3) physical network GOutput 119866119871(1) 119873119903 997888rarr 1198811 119871119872119860119883 = L1(2) for 119881119871119872119860119883 = 0 119873119898 = 119873119886 do(3) for each119873119899 isin 119866 do(4) if 119873119898 larrrarr 119873119899 ampamp 119873119899 notin sum1198711198721198601198831 119881119883(5) 119873119899 997888rarr 119881119871119872119860119883+1 (6) else if 119873119898 larrrarr 119873119899 ampamp 119873119899 isin sum1198711198721198601198831 119881119883ampamp119873119899 =119873119886(7) 119873119899 997888rarr 119881119871119872119860119883+1 (8) end if(9) end for(10) 119871119872119860119883 ++(11) end for(12) for LX le 119871119872119860119883 do(13) for 119873119898 isin 119881119883 do(14) 119873119898 997888rarr 119881(1198831)(15) 119871119898119872119860119883 = 1198711198981(16) for 119881(119883119871119898119872119860119883) = 0 do(17) if 119873119899 isin 119881119883ampamp119873119898 larrrarr 119873119899ampamp119873119899 notin sum1198711198981198721198601198831 119866119883L(18) 119873119899 997888rarr 119881(119883119871119898119872119860119883)(19) end if(20) 119871119898119872119860119883 ++(21) end for(22) end for(23) end for(24) return 119866119871
Algorithm 2 Network layered processing
Input (1) SFC request(2) 119866119871(3) bool direction(4) X 119873119871 isin 119881119883Output the node119873119862 which has the minimum value of 120575(1) Temp = +infin(2) int i= 0(3) if direction is true(4) i= X-1(5) end if(6) if direction is false(7) i= X+1(8) end if(9) for 119873119898 isin 119881119894 do(10) if 119873119898 larrrarr 119873119871(11) if 119861119904119894 119898 gt 119861119903119894 119898ampamp119861119904119890 119898 gt 119861119903119890 119898 ampamp 119862119904 119898 gt 119862119903(12) Compute 120575 according to Equation (7)(13) if 120575 lt Temp(14) Temp = 120575(15) 119873119862 =119873119898(16) end if(17) end if(18) end if(19) end for(20) return 119873119862
Algorithm 3 Evaluation of related nodes
8 Security and Communication Networks
N1
N2
N3
N4 N6
N5
N8 N10
N9
N7
(a) Physical network
L1 V1 L3 V3L2 V2 L4 V4 L5 V5 L6 V6
N2
N3
N4
N5
N6
N9
N7
N8
N9
N10
N9N1
N4 N3N3 N4
6(21)3
6(22)3
6(21)4
6(22)4
(b) Processing procedure of layering
Figure 3 An example of a layered physical network
6 and there is an inner layer in layer L2The total sum is 7 sothis network topology can satisfy only an SFC request whoselength is no more than 7 An SFC request whose number ofVNFs ismore than 7 is too large for this network However aslong as the number of VNFs is not greater than the numberof nodes in the network our OSFCD-LSEM algorithm cantry to find more paths to deploy the longer SFC It may needadditional time and capacity of bandwidth resources becausethere are too many VNFs that need to be deployed Theproposed Algorithm 3 can search nodes in both positive andnegative directions When addressing an SFC that is longerthan the abovementioned sum Algorithm 3 commonly findsa suitable node in the layer119881119871119860+1 instead of in the upper layer119881119871119860minus1 then the algorithm can increase the length of the path(119871119875) However an extreme situation should be consideredsuch as when the node is in the last layer L119871119872119860119883 For thissituation our algorithm will find a node in the upper layer119881119871119872119860119883minus1 and run Algorithm 2 again
Finally we must evaluate the nodes from each layer ofthe layered network and select some nodes to map the VNFsAlgorithm 3 uses the abovementioned strategy to find a pathfrom 119873119886 to 119873119903 to host an SFC request and satisfy the userdemand Algorithm 3 selects the nodes from each layer using(7) and (8) The selected node must directly link to thenode in the next layer VN and must have sufficient resources
to deploy the VNFs and satisfy the corresponding functionrequirements
120575 = min(119861119904119894 119898 minus 119861119903119894 V119899119891119894119895119861119904119894 119898 119861119904119890 119898 minus 119861119903119890 V119899119891119894119895119861119904119890 119898 )
times 119862119904 119898 minus 119862119903 V119899119891119894119895119862119904 119898(7)
119862119904 119898 = 119862119905119900119905119886119897 119898 minus sum119878119894isin119878119865119862 119900119899119897119894119899119890
sumV119899119891119894119895isin119878119894
119862119903 V119899119891119894119895 times 120572V119899119891119894119895 119898 (8)
In (7) we use 120575 to denote the nodersquos appropriateness forthe VNF in the user request 119861119904119894 119898 is the idle bandwidth ofall links that connect node Nm with the nodes in the nextlayer 119881119873 and 119861119903119894 V119899119891119894119895 is the requested bandwidth resourcefrom vnf ij to the vnf ij+1 where V119899119891119894j is the j-th VNF in theSi (ie the i-th SFC request) 119861119904119890 119898 is the idle bandwidthof the path that connects node Nm with the nodes in theupper layer 119881119880 and 119861119903119890 V119899119891119894119895 is the requested bandwidthresource for making vnf ij connect with the last VNF 119862119904 119898represents the idle computing resources of node Nm and119862119903 V119899119891119894119895 represents the requested computing resources of thevnf ij In (8) Ctotal m represents the total computing resourcesof node Nm SFConline is the set of online SFC requests and
Security and Communication Networks 9
(a) US-Net topology (b) China-Net topology
Figure 4 Two real network topologies used in our simulations
120572V119899119891119894119895 119898 denotes whether vnf ij is deployed on node Nm Wecan evaluate whether a physical node is suitable to deploythe corresponding VNF according to the value of 120575 Afterphysical node evaluation we select the nodewith theminimalvalue of 120575 to hold the corresponding VNF
5 Simulation Results and Analysis
To evaluate the performance of our algorithm we compareour algorithm with two existing algorithms ie closed-loopwith critical mapping feedback (CCMF) [17] and key-VNFdeploy first (KVDF) [38] CCMF deploys the VNFs that havemore resource requirement priorities to optimize the totalconsumption of bandwidth resources KVDF focuses on therelation between different VNFs in the same SFC and therelation among different SFCs According to the relation andinfluence KVDF prioritizes the deployment of the key-VNFand key-SFC
To evaluate the performance of algorithms in differentnetwork scenarios we evaluate the performance of comparedalgorithms in moderate-scale and large-scale networks theUS-Network [50] (shown in Figure 4(a)) and the China-Network [51] (shown in Figure 4(b)) We use GT-ITM [52] togenerate the moderate-scale and large-scale network topolo-gies In moderate-scale network topologies there are approx-imately 100 nodes and 400 links In large-scale networktopologies there are approximately 300 nodes and 1000 linksTheUS-Net topology has 24 nodes and 43 links Additionallythe China-Net topology has 55 nodes and 103 links In thesenetwork topologies the bandwidth resource of all links isuniformly distributed at 100sim200 units and the computingresource of all nodes is set as 10 units
In our simulations we set the following parametersaccording to the existing work [49] The computer randomlygenerates user requests with lengths (ie Ls) from 5 to 14 andthese online SFC requests arrive as a Poisson process For eachphysical network and for a given 119871 119904 we randomly generate10000 SFC requests with the request client and destinationclient nodes which are randomly assigned to the physicalnodesThe computing resource demand of each VNF followsa uniform distribution U (1 3) and the bandwidth resource
demand of each virtual link follows a uniform distribution U(20 50)
With the increasing number of users and SFC requeststhe deployment of SFC in a static network becomes increas-ingly challenging Thus the network scalability must beimproved For a network-aware scaling strategy it is betterto extend the network instead of changing the network Innetwork G we define the evaluation of information 119866119878 in
119866119878 = 119871119872119860119883sum119883=1
sum119873119894isin119881119883
(119862119904 + 119861119904119894 + 119861119904119890) (9)
The OSFCD-LSEM algorithm layers the physical net-work obtains the layer which has minimum resourcesanalyzes its inner layer information and obtains the ldquoweakrdquonodes or links that influence the capacity of the networkThen the OSFCD-LSEM algorithm extends their resourcesto secure a more robust network
We obtained the simulation results by averaging theresults of multiple simulations We used an Ubuntu virtualmachine running on a computer with a 37 GHz Intel Corei3-4170 and 4 GB of RAM to run the simulations And thealgorithms were coded in Java programming language
From Figure 5 we can see the simulation results ofour OSFCD-LSEM algorithm in a moderate-scale networkFigure 5(a) shows the information of the physical networkWe can see that L8 limits the network capacity and will affectthe deployment of SFC Figure 5(b) shows the information ofthe physical nodes in L8 In L8 node-67 has the minimalamount of bandwidth and node-72 has the minimal amountof computing resources Both node-67 and node-72 are theldquoweak pointsrdquo in the physical network If we can increasetheir corresponding resources the capacity of the physicalnetwork will be enhanced For network operators increasingthe corresponding resources to the corresponding nodes andlinks is necessary and highly beneficial moreover they canalso obtain a more robust network
Figure 6 shows the simulation results on the US-Netnetwork Figure 6(a) shows the information about each layerin the overall network and we find that L4 may be theldquoweak pointrdquo of the physical network because of the lackof computing and bandwidth resources Figure 6(b) shows
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
2 Security and Communication Networks
Physical Resources
Virtualized Resource Pool
VNF Deployment and Chaining
VNF1-NAT
VNF2-IDS
VNF4-Firewall
VNF3-Load
Balance
SFC
User User
NAT IDS FirewallLoad Balance
Server Server
Server Server
VM
VM
Server Server
Server Server
VM
VMVM
VM
VM
VM
Figure 1 SFC organization by virtualizing commercial servers
traditional NFs VNFs increase policy compliance capabil-ities [29] and promote the emergence of more novel tech-niques and the development of NFV for social IoT applica-tion
NFV can bring additional benefits to social IoT ser-vice providers ie optimize their operational expenditure(OPEX) and capital expenditure (CAPEX) [11 30ndash32] andimprove the quality of service (QoS) by decreasing latencyreducing bandwidth consumption and increasing adaptationNFV introduces many benefits to both users and social IoTservice providers but there are still many research limitationsthat need to be dealt with For example the bandwidthconsumption of the SFC is an important research topic
Bandwidth consumption represents the resource cost ofcommunication among usersclients [33 34] The networkoperator seeks to achieve low bandwidth consumption tocomplete the deployment of the current SFC request and saveresources for the next SFC request to improve the accommo-dating capacity of the network [35]With an increasing diver-sity of service requirements and additional pressure frommassive information transmission bandwidth resources havebecome progressively scarce [36] Furthermore low band-width consumption can provide the clientuser a goodexperience due to the excellent performance of the networkoperator In previous research several algorithms have aimedto reduce bandwidth consumption In [17] the authors
Security and Communication Networks 3
focused on bandwidth consumption optimization and pro-vided a framework for their studied problem Thus insteadof the existing hardware environment targeted researchof NFV can produce increased benefits and reduce theenergy and space consumption of various middle boxes[19 37]
In this paper we study the problem of how to provisionsocial IoT oriented SFC requests while minimizing the band-width resource consumption as well supporting the securityand privacy requirements The studied problem has beenproven to be an NP-hard problem [35 37 38]
Thus we propose an efficient algorithm called OSFCD-LSEM with layered strategies for physical networks It notonly efficiently solves the SFC deployment problem but alsomakes the physical networkmore robust Our OSFCD-LSEMalgorithm evaluates the nodes in the physical network tooptimize the bandwidth consumption and thus achievesa higher acceptance ratio and shorter response time (ielatency) for user demands than the exiting work In additionwe can make the physical network more robust using ourOSFCD-LSEM algorithm The main contributions of thispaper are as follows
(i) We develop a model to evaluate the physical networkand extend it precisely to its ldquoweak pointsrdquo and makethe physical network more robust
(ii) We propose an efficient algorithm (ie OSFCD-LSEM) for social IoT oriented SFC deployment that isbased on the layered strategies of a physical networkand evaluation of a physical network node to optimizethe consumption of bandwidth resources and guaran-tee the security and privacy
(iii) We conduct extensive simulations to verify and eval-uate our algorithm in this paper The results showthat our proposed algorithm can efficiently addressthe online SFC deployment problem
The remainder of this paper is organized as followsSection 2 provides an overview of related works Section 3provides the problem descriptions and formulation InSection 4 we describe our SFC deployment algorithm basedon the layered strategies of physical networks and eval-uation of physical network nodes Section 5 shows thesimulation results and analysis Section 6 summarizes thispaper
2 Related Work
NFV aims to satisfy client demands with minimal resourceconsumptions (eg bandwidth consumption and computingcore consumption) and high performance (eg low latencyand high throughput) [20] An increasing number of studieshave focused on the more efficient deployment of social IoToriented SFC requests in various scenarios
The research in [39] focused on the suitability of differentarchitectures of data center for a resilient SFC and theplacement of VNFs with high availability constraints In[40] the joint VNF placement and path selection problemwere studied The authors proposed a chain deployment
algorithm to balance the chain length and reuse factor itstarget was to serve as many demands as possible underlimited resources The research in [41] designed a heuristicalgorithm to address the NFV-RA problem in a coordinatedmanner by splitting the SFC request In [42] the authormade a BCMP mixed queuing network to replace the SFCmodel and proposed the convex optimization problem toshorten the acceptance interval of the service chain Thesimulation results shown the algorithm in [42] can effec-tively adapt the resource usage to the network dynamicsand serve more demands than other existing algorithmsThe authors of [43] studied SFC deployment in a multi-domain network and proposed a vertex-centric distributedcomputing algorithm to find all feasible SFC deploymentmethods of user requests in the distributed orchestrationframework then the algorithm selects the most appropriatescheme to deploy the SFC to achieve efficient performanceIn [44] the researcher discussed the phenomenon that theVNFs may sprawl across the network due to inefficientmapping during the SFC orchestration process and designedan efficient greedy heuristic algorithm to solve the problemThe author in [45] studied the service availability constraintsin a data center and designed a heuristic algorithm to deploythe SFC In the situation of distributed NFVs the authors in[46] studied the security level of a network security frame-work The authors proposed a new optimization algorithmto avoid a single point of failure in a bottleneck problemand obtained reasonable performance relative to that of acommon device Li et al [47] presented a real-time resource-distributing system for NFV which integrates timing analysiswith other algorithms such as timing abstraction SFC con-solidation and linear programming algorithm to efficientlydistribute network resources while considering latency con-straints
Many researchers have studied the deployment of thesocial IoT oriented VNF or SFC and a few of them haveattached importance to the total bandwidth consumptionYe et al [17] focused on the joint topology design and SFCdeployment problem to minimize the total consumption ofbandwidth resources Their main idea was to reduce band-width usage and cost by prioritizing VNF deployment In[48] the authors proposed and analyzed two deploymentstrategies HSD and VSD whose performance was analyzedin terms of cost They claimed that HSD had better per-formance in terms of an even distribution of the load overall servers access links and ToR switches The authors in[49] combined NFV with cloud computing and designeda bandwidth-guaranteed SFC-placing algorithm then theauthors tested the performance of their algorithm in a datacenter The simulation results showed that the heuristicalgorithm had excellent performance In [38] the authorsdesigned a forecast-assisted online SFC deployment algo-rithm that included the prediction of future VNF require-ments The simulation results showed that the algorithm in[38] could reduce the blocking probability of SFC requestsand effectively improve the profit of the service provider fromSFC deployment
Overall bandwidth is the most prominently expensiveresource that directly affects the number of demands that the
4 Security and Communication Networks
Client Client
Client
Client
90
f150
f270
f380 90
N4N5N3
N1
N2 N6
N7
N8
70
7080
8050
(a)
Client
Client
8070
70
N4N5N3
N1
N2 N6
N7
N8
Client Clientf150
f270
f380 90
(b)
Figure 2 Two examples of SFC deployment
network can satisfy online and the capacity of the physicalnetworkThus in this research we focused on the bandwidthconsumption and designed an efficient algorithm to provisionsocial IoT oriented SFC requests
3 Problem Formulation
In this paper we studied the online social IoT oriented SFCdeployment problem We considered a situation in whicheach SFC request has a pair of start and terminal nodesattached to the given physical network nodes and a specificsequence of VNFs that make up continuous functions Wehave to deploy these VNFs onto the corresponding physicalnodes and then organize the VNFs to form an SFC Toreduce the bandwidth resource consumption we need to usefewer nodes to shorten the length of the SFC as much aspossible
A user request can be denoted as 119878 = (119865S 119864S)where 119865S = 1198911 1198912 119891m is the set of VNFs and119864S = 1198901 1198902 119890q represents the virtual link after SFCdeployment The real physical underlying network can bemodeled as G = (N L) where G is an undirected weightedgraph 119873 = 11987311198732 119873y is the set of physical nodesand 119871 = 1198711 1198712 119871k is the set of real links inthe physical underlying network We use 119875 to denote thephysical path that holds the SFC request We define 119862119879119861 todenote the total bandwidth consumption which is definedin
119862119879119861 = sum119871 119894 119890119895isin119875 119878
119862119871 119894 119890119895119861 (1)
where119871 119894 119890119895 denotes that virtual link 119890119895 is deployed on the linkLi and 119862119871 119894 119890119895119861 is the bandwidth consumption of virtual link 119890119895which is deployed on the linkLi Wedefine119877119873119894119862 as the availablecomputing resources of the physical node119873119894 and 119862119873119894 119891119895119873 as thecomputing resource requirements of VNF 119891119895 which wouldbe deployed on the node Ni 119877119871 119894119861 is the available bandwidthresources of the physical link 119871119894
To deploy an SFC we must map the VNFs to some nodeand the virtual links 119864S to some real links in the physicalnetwork the path 119875 119878 which would hold the SFCs musthave sufficient nodes and computing resources to deploythe corresponding VNFs and the physical links must havesufficient available bandwidth for communication amongthose VNFs In addition the available bandwidth resourcesmust satisfy the requirements of the 119864S in the correspondinguser request Similar to [10 20] this paper assumes thateach VNF from the same SFC needs to use different physicalnetwork nodes Then the number of VNFs (denoted as Fs)should be less than the number of nodes of physical pathP 119878(denoted by 119875 119878) Then the problem of SFC deploymentcan be formulated as follows
min sum119871 119894 119890119895isin119875 119878
119862119871 119894 119890119895119861119904119905 forall119871 119894 119890119895 isin 119875 119878
119877119871 119894119861 minus 119862119871 119894 119890119895119861 ge 0forall119873119894 119891119895 isin 119875 119878119877119873119894119862 minus 119862119873119894 119891119895119873 ge 0119875 119878 minus 119865119904 ge 0
(2)
Formula (2) is used to model the SFC deployment prob-lem which can minimize the bandwidth consumptions whilefinishing SFC deployment There must be sufficient availablecomputing resources to deploy corresponding SFCs and thebandwidth must be sufficient to satisfy the communicationdemands among the corresponding VNFs In addition theremust be sufficient nodes in the physical path P to deploy theVNFs of the SFC request
Figure 2 shows two different examples of provisioningan SFC request both of which successfully deploy the SFCrequest and satisfy the clientsrsquo demands Nodes N1 and N8are the requested client node and destination client noderespectively There are three VNFs (ie f 1 f 2 and f 3) inthe SFC request Client node N1 needs to transmit 50 units
Security and Communication Networks 5
of traffic to VNF f 1 Between VNFs f 1 and f 2 70 unitsof information need to be transmitted Between VNFs f 2and f 3 80 units of traffic need to be transmitted Thereis a 90-unit transmission demand from VNF f 3 to thedestination client node N8 As shown in Figure 2(a) wedeploy the VNFs f 1 f 2 and f 3 onto the physical nodesN2 N5 and N6 respectively Then we find the path P =1198731-11987321198732-11987331198733-11987351198735-11987371198737-11987361198736-1198738 to hostthe SFC In this deployment scheme the total bandwidthconsumption is 440 units Although this approach success-fully deploys the SFC request it wastes bandwidth resourcesIn Figure 2(b) the VNFs f 1 f 2 and f 3 are deployed ontothe physical nodes N1 N6 and N8 respectively Then wefind a short path P = 1198731-11987321198732-11987361198736-1198738 (shown asthe red line in Figure 2(b)) which can also deploy all VNFsfrom the user requests However the total consumptionof bandwidth resources of this scheme is only 220 unitswhich is approximately half of the consumption of thedeployment in Figure 2(a) In Figure 2(b) path P is theshortest path to deploy this SFC request and this schemecan reduce more bandwidth consumption Mapping theSFC to the path in Figure 2(b) the service network candeploy more SFC with other links which is impossible inFigure 2(a)
Therefore different deployment schemes significantlyaffect the bandwidth consumptions and thus affect the avail-able network capacity An efficient algorithm is important andurgently needed to better deploy SFC requests and reducebandwidth consumption
4 Algorithm Design
Since the optimal SFC deployment problem is NP-hard [3537] we design an efficient algorithm for solving our researchproblem in this section Our proposed algorithm employsthe layered strategies of physical networks and evaluationof physical network nodes to optimize the consumption ofbandwidth resources which is denoted by OSFCD-LSEMThe basic idea of OSFCD-LSEM is to find the shortest pathsto deploy SFC requests while saving as many bandwidthresources as possible to satisfy more user requests Whena user demand arrives the OSFCD-LSEM algorithm beginsto handle it First calling Algorithm 2 the OSFCD-LSEMalgorithm layers the physical underlying network and obtainsthe information from the nodes and links in each layer ofthe physical network Then it calls Algorithm 3 to performan evaluation of nodes in the network and select some nodesthat are most suitable to host the VNFs of this SFC requestFinally it connects the deployed VNFs by using a shortestpath to fulfill an SFC By layered strategies and selection ofthe nodes for VNF deployment the OSFCD-LSEMalgorithmcan deploy VNFs in the most suitable nodes and deploythe SFC requests in appropriate simple paths to save morebandwidth resources The physical path must contain thestart node Nr and the terminal node Na Furthermore thenodes in the physical path must have sufficient resources tohost the VNFs of the user request
In the proposed OSFCD-LSEM algorithm we providesome definitions of variables 119866119871 is the physical network after
layering and 119881119883 is used to denote the set of nodes in the X-th layer The X-th layer is denoted as LX 119866119883119871 is the innerlayered network in the X-th layer (LX) and we used LY todenote the Y-th layer in 119866119883119871 119881119894(119883119884) is the set of nodes in theLY of 119866119883119871 which includes the node Ni 119864119883 is the set of linksthat connect the nodes from LX and LX-1 119864119894(119883119884) denotesthe corresponding links that connect the nodes in the LY-1about nodeNi and 119871119894119872119860119883 is the correspondingmaximal layerof node Ni LMAX is the layer number of GL 119873119879 is the nodenumber of the physical network and 119871119879 is the link numberof the physical network G The OSFCD-LSEM algorithm isshown in Algorithm 1
Next Algorithm 2 is responsible for handling hierarchicalphysical networks in our algorithm Algorithm 2 layers theentire network to obtain the layering information of thenodes and links in the network and outputs the results toother algorithms Therefore Algorithm 2 is the basis of ourSFC request deployment scheme and physical network nodeevaluation The physical network layering can be formulatedas follows
119866119871 =119871119872119860119883sum119883=1
(119881119883 119864119883) +119871119872119860119883sum119883=2
119866119883119871 (3)
119866119883119871 = sum119881119894isin119881119883
119871119894119872119860119883sum119884=1
(119881119894(119883119884) 119864119894(119883119884)) (4)
119871119872119860119883sum119883=1
119881119883 minus 119873119879 ge 0 (5)
119871119872119860119883sum119883=2
119864119883 +119871119872119860119883sum119883=2
sum119881119894isin119881119883
119871119894119872119860119883sum119884=2
119864119894(119883119884) minus 119871119879 = 0 (6)
In (3) two parts make up 119866119871 one part is the inner layernetwork 119866119883119871 about the X-th layer (LX) and the other partis the overall layered network The layering process beginsfrom the request node 119873119903 thus 1198811 = 119873119903 1198641 = 0 and1198661119871 = 0 Equation (4) shows that to make 119866119871 closer to thephysical network G each layer except layer L1 should obtainits inner layer node-related information The OSFCD-LSEMalgorithmcanobtain amore precise evaluation of the physicalnetwork and more efficiently deploy the corresponding SFCrequests In (5) the equation describes that all nodes in thelayered physical network must be in the corresponding layerIn (6) the equation shows that each link should be in thecorresponding layer or inner layer
In Figure 3 we give an example of layering a networkFigure 3(a) shows the topology of a physical network andit has a total of 10 nodes Figure 3(b) provides the detailedprocessing procedure for layering the network topology Weassume that the start node 119873119903 in the request is node N1 andthat the terminal node 119873119886 in the request is node N9 Firstour algorithm places the start node N1 into L1 (N1 is theonly node in V1 of L1) and places nodes N2 N3 and N4into L2 because they directly have links to node N1 Thenour algorithm places nodesN5N6 andN9 all of which havelinks to some nodes in the L2 (ie nodes N2 N3 and N4)
6 Security and Communication Networks
Input (1) SFC request (2) physical network 119866Output SFC deployment scheme(1) Receive a user request(2) Initialize Path = [](3) 119873119886 997888rarr Path119873119871 =119873119886(4) Layer the topology Algorithm 2(119873119903119873119871 119866)(5) Get 119871119860 the layers that destination client belongs to(6) while 119871119878 gt max(119871119860) + sum119871119872119860119883119883=1 max(119871119894119872119860119883) forall119873119894 isin 119881119883 do(7) if max119871119860 = 119871119872119860119883(8) 119873119879119864119872119875 = Algorithm 3( true max119871119860)(9) 119873119879119864119872119875 997888rarr Path(10) 119873119871 =119873119879119864119872119875(11) else(12) 119873119879119864119872119875 = Algorithm 3( false max119871119860)(13) 119873119879119864119872119875 997888rarr Path(14) 119873119871=119873119879119864119872119875(15) end if(16) 119871119878 = 119871119878 - 1(17) VNF 997888rarr 119873119879119864119872119875(18) Algorithm 2(119873119903119873119871 119866)(19) Update 119871119860(20) end while(21) if 119871119878 lemax(119871119860)(22) Select min LX isin 119871119860 ampamp LX gt 119871119878(23) while 119873119903 notin Path do(24) 119873119879119864119872119875 = Algorithm 3 ( true LX)(25) 119873119879119864119872119875 997888rarr Path(26) 119873119871 =119873119879119864119872119875(27) LX = LX - 1(28) VNF997888rarr 119873119879119864119872119875(29) end while(30) end if(31) SFC deployment scheme is Path
Algorithm 1 OSFCD-LSEM algorithm
into L3 (we specify that all nodes can only belong to onelayer except the terminal node N9 thus node N4 cannot bepart of L3 even though it connects with node N3 which isin L2)
In our layered network except for the destination clientnode N9 the nodes in one layer must have links to the nodesin the previous layer Thus nodes N7 N8 and N10 directlyhave links to the nodes in L3 (ie nodes N5 N6 and N9)whereas N10 connects only with destination client node N9among the three nodes in L3 NodeN10 should not be placedin layer L4 Thus we place only nodes N7 and N8 into L4Nodes N9 and N10 connect with nodes N7 and N8 in L4thus we place nodes N9 and N10 in layer L5 and becauseN9 connects with node N10 our algorithm places node N9in L6 When ten nodes in G belong to the correspondinglayers the overall network-layered processing finishes Thusthe OSFCD-LSEM algorithm can guarantee finding a pathwithout loops For each LX we also need to layer and obtainthe inner layer 119866119883119871 In the example shown by Figure 3(b) thelayer L2 has an inner layer 1198662119871 that includes two layers For1198662119871 in Figure 3(b) each layer X le LMAX and each node Ni isinVX should be set as the start node 119873119903 let 119873119886 = 0 Then we
obtain their inner layer information In 1198662119871 both 1198711198733119872119860119883 and1198711198734119872119860119883 are equal to 2 whereas 1198711198732119872119860119883 is equal to 1Overall the physical network is layered into six layers
with two inner layer networks associated with nodes N3 andN4The start node119873119903 is the only one in the first layer and theterminal node119873119886 is in three layers including L3 L5 and L6Thus we obtain three paths that can be used between119873119903 and119873119886 119871119875 is used to denote the length of path which is equal tothe number of VNFs that the path can hold LS is the lengthof an SFC which is equal to the VNF number of the SFCrequest
Algorithm 3 evaluates the nodes in the physical networkand selects a node which is most suitable to deploy therequired VNF After obtaining the layering informationAlgorithm 3makes the decision regarding whether themulti-ple links can satisfy the user request When themaximal layernumber in 119871119860 of the terminal nodes 119873119886 which is obtainedfrom the sumof all inner layers is smaller than the SFC length119871119878 in the user request the physical network cannot satisfy theuser request For example if we need to deploy an SFC intothe physical network in Figure 3(a) the start and terminalnodes are N1 and N9 The maximum layer number in 119871119860 is
Security and Communication Networks 7
Input (1) 119873119903 (2) 119873119886 (3) physical network GOutput 119866119871(1) 119873119903 997888rarr 1198811 119871119872119860119883 = L1(2) for 119881119871119872119860119883 = 0 119873119898 = 119873119886 do(3) for each119873119899 isin 119866 do(4) if 119873119898 larrrarr 119873119899 ampamp 119873119899 notin sum1198711198721198601198831 119881119883(5) 119873119899 997888rarr 119881119871119872119860119883+1 (6) else if 119873119898 larrrarr 119873119899 ampamp 119873119899 isin sum1198711198721198601198831 119881119883ampamp119873119899 =119873119886(7) 119873119899 997888rarr 119881119871119872119860119883+1 (8) end if(9) end for(10) 119871119872119860119883 ++(11) end for(12) for LX le 119871119872119860119883 do(13) for 119873119898 isin 119881119883 do(14) 119873119898 997888rarr 119881(1198831)(15) 119871119898119872119860119883 = 1198711198981(16) for 119881(119883119871119898119872119860119883) = 0 do(17) if 119873119899 isin 119881119883ampamp119873119898 larrrarr 119873119899ampamp119873119899 notin sum1198711198981198721198601198831 119866119883L(18) 119873119899 997888rarr 119881(119883119871119898119872119860119883)(19) end if(20) 119871119898119872119860119883 ++(21) end for(22) end for(23) end for(24) return 119866119871
Algorithm 2 Network layered processing
Input (1) SFC request(2) 119866119871(3) bool direction(4) X 119873119871 isin 119881119883Output the node119873119862 which has the minimum value of 120575(1) Temp = +infin(2) int i= 0(3) if direction is true(4) i= X-1(5) end if(6) if direction is false(7) i= X+1(8) end if(9) for 119873119898 isin 119881119894 do(10) if 119873119898 larrrarr 119873119871(11) if 119861119904119894 119898 gt 119861119903119894 119898ampamp119861119904119890 119898 gt 119861119903119890 119898 ampamp 119862119904 119898 gt 119862119903(12) Compute 120575 according to Equation (7)(13) if 120575 lt Temp(14) Temp = 120575(15) 119873119862 =119873119898(16) end if(17) end if(18) end if(19) end for(20) return 119873119862
Algorithm 3 Evaluation of related nodes
8 Security and Communication Networks
N1
N2
N3
N4 N6
N5
N8 N10
N9
N7
(a) Physical network
L1 V1 L3 V3L2 V2 L4 V4 L5 V5 L6 V6
N2
N3
N4
N5
N6
N9
N7
N8
N9
N10
N9N1
N4 N3N3 N4
6(21)3
6(22)3
6(21)4
6(22)4
(b) Processing procedure of layering
Figure 3 An example of a layered physical network
6 and there is an inner layer in layer L2The total sum is 7 sothis network topology can satisfy only an SFC request whoselength is no more than 7 An SFC request whose number ofVNFs ismore than 7 is too large for this network However aslong as the number of VNFs is not greater than the numberof nodes in the network our OSFCD-LSEM algorithm cantry to find more paths to deploy the longer SFC It may needadditional time and capacity of bandwidth resources becausethere are too many VNFs that need to be deployed Theproposed Algorithm 3 can search nodes in both positive andnegative directions When addressing an SFC that is longerthan the abovementioned sum Algorithm 3 commonly findsa suitable node in the layer119881119871119860+1 instead of in the upper layer119881119871119860minus1 then the algorithm can increase the length of the path(119871119875) However an extreme situation should be consideredsuch as when the node is in the last layer L119871119872119860119883 For thissituation our algorithm will find a node in the upper layer119881119871119872119860119883minus1 and run Algorithm 2 again
Finally we must evaluate the nodes from each layer ofthe layered network and select some nodes to map the VNFsAlgorithm 3 uses the abovementioned strategy to find a pathfrom 119873119886 to 119873119903 to host an SFC request and satisfy the userdemand Algorithm 3 selects the nodes from each layer using(7) and (8) The selected node must directly link to thenode in the next layer VN and must have sufficient resources
to deploy the VNFs and satisfy the corresponding functionrequirements
120575 = min(119861119904119894 119898 minus 119861119903119894 V119899119891119894119895119861119904119894 119898 119861119904119890 119898 minus 119861119903119890 V119899119891119894119895119861119904119890 119898 )
times 119862119904 119898 minus 119862119903 V119899119891119894119895119862119904 119898(7)
119862119904 119898 = 119862119905119900119905119886119897 119898 minus sum119878119894isin119878119865119862 119900119899119897119894119899119890
sumV119899119891119894119895isin119878119894
119862119903 V119899119891119894119895 times 120572V119899119891119894119895 119898 (8)
In (7) we use 120575 to denote the nodersquos appropriateness forthe VNF in the user request 119861119904119894 119898 is the idle bandwidth ofall links that connect node Nm with the nodes in the nextlayer 119881119873 and 119861119903119894 V119899119891119894119895 is the requested bandwidth resourcefrom vnf ij to the vnf ij+1 where V119899119891119894j is the j-th VNF in theSi (ie the i-th SFC request) 119861119904119890 119898 is the idle bandwidthof the path that connects node Nm with the nodes in theupper layer 119881119880 and 119861119903119890 V119899119891119894119895 is the requested bandwidthresource for making vnf ij connect with the last VNF 119862119904 119898represents the idle computing resources of node Nm and119862119903 V119899119891119894119895 represents the requested computing resources of thevnf ij In (8) Ctotal m represents the total computing resourcesof node Nm SFConline is the set of online SFC requests and
Security and Communication Networks 9
(a) US-Net topology (b) China-Net topology
Figure 4 Two real network topologies used in our simulations
120572V119899119891119894119895 119898 denotes whether vnf ij is deployed on node Nm Wecan evaluate whether a physical node is suitable to deploythe corresponding VNF according to the value of 120575 Afterphysical node evaluation we select the nodewith theminimalvalue of 120575 to hold the corresponding VNF
5 Simulation Results and Analysis
To evaluate the performance of our algorithm we compareour algorithm with two existing algorithms ie closed-loopwith critical mapping feedback (CCMF) [17] and key-VNFdeploy first (KVDF) [38] CCMF deploys the VNFs that havemore resource requirement priorities to optimize the totalconsumption of bandwidth resources KVDF focuses on therelation between different VNFs in the same SFC and therelation among different SFCs According to the relation andinfluence KVDF prioritizes the deployment of the key-VNFand key-SFC
To evaluate the performance of algorithms in differentnetwork scenarios we evaluate the performance of comparedalgorithms in moderate-scale and large-scale networks theUS-Network [50] (shown in Figure 4(a)) and the China-Network [51] (shown in Figure 4(b)) We use GT-ITM [52] togenerate the moderate-scale and large-scale network topolo-gies In moderate-scale network topologies there are approx-imately 100 nodes and 400 links In large-scale networktopologies there are approximately 300 nodes and 1000 linksTheUS-Net topology has 24 nodes and 43 links Additionallythe China-Net topology has 55 nodes and 103 links In thesenetwork topologies the bandwidth resource of all links isuniformly distributed at 100sim200 units and the computingresource of all nodes is set as 10 units
In our simulations we set the following parametersaccording to the existing work [49] The computer randomlygenerates user requests with lengths (ie Ls) from 5 to 14 andthese online SFC requests arrive as a Poisson process For eachphysical network and for a given 119871 119904 we randomly generate10000 SFC requests with the request client and destinationclient nodes which are randomly assigned to the physicalnodesThe computing resource demand of each VNF followsa uniform distribution U (1 3) and the bandwidth resource
demand of each virtual link follows a uniform distribution U(20 50)
With the increasing number of users and SFC requeststhe deployment of SFC in a static network becomes increas-ingly challenging Thus the network scalability must beimproved For a network-aware scaling strategy it is betterto extend the network instead of changing the network Innetwork G we define the evaluation of information 119866119878 in
119866119878 = 119871119872119860119883sum119883=1
sum119873119894isin119881119883
(119862119904 + 119861119904119894 + 119861119904119890) (9)
The OSFCD-LSEM algorithm layers the physical net-work obtains the layer which has minimum resourcesanalyzes its inner layer information and obtains the ldquoweakrdquonodes or links that influence the capacity of the networkThen the OSFCD-LSEM algorithm extends their resourcesto secure a more robust network
We obtained the simulation results by averaging theresults of multiple simulations We used an Ubuntu virtualmachine running on a computer with a 37 GHz Intel Corei3-4170 and 4 GB of RAM to run the simulations And thealgorithms were coded in Java programming language
From Figure 5 we can see the simulation results ofour OSFCD-LSEM algorithm in a moderate-scale networkFigure 5(a) shows the information of the physical networkWe can see that L8 limits the network capacity and will affectthe deployment of SFC Figure 5(b) shows the information ofthe physical nodes in L8 In L8 node-67 has the minimalamount of bandwidth and node-72 has the minimal amountof computing resources Both node-67 and node-72 are theldquoweak pointsrdquo in the physical network If we can increasetheir corresponding resources the capacity of the physicalnetwork will be enhanced For network operators increasingthe corresponding resources to the corresponding nodes andlinks is necessary and highly beneficial moreover they canalso obtain a more robust network
Figure 6 shows the simulation results on the US-Netnetwork Figure 6(a) shows the information about each layerin the overall network and we find that L4 may be theldquoweak pointrdquo of the physical network because of the lackof computing and bandwidth resources Figure 6(b) shows
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 3
focused on bandwidth consumption optimization and pro-vided a framework for their studied problem Thus insteadof the existing hardware environment targeted researchof NFV can produce increased benefits and reduce theenergy and space consumption of various middle boxes[19 37]
In this paper we study the problem of how to provisionsocial IoT oriented SFC requests while minimizing the band-width resource consumption as well supporting the securityand privacy requirements The studied problem has beenproven to be an NP-hard problem [35 37 38]
Thus we propose an efficient algorithm called OSFCD-LSEM with layered strategies for physical networks It notonly efficiently solves the SFC deployment problem but alsomakes the physical networkmore robust Our OSFCD-LSEMalgorithm evaluates the nodes in the physical network tooptimize the bandwidth consumption and thus achievesa higher acceptance ratio and shorter response time (ielatency) for user demands than the exiting work In additionwe can make the physical network more robust using ourOSFCD-LSEM algorithm The main contributions of thispaper are as follows
(i) We develop a model to evaluate the physical networkand extend it precisely to its ldquoweak pointsrdquo and makethe physical network more robust
(ii) We propose an efficient algorithm (ie OSFCD-LSEM) for social IoT oriented SFC deployment that isbased on the layered strategies of a physical networkand evaluation of a physical network node to optimizethe consumption of bandwidth resources and guaran-tee the security and privacy
(iii) We conduct extensive simulations to verify and eval-uate our algorithm in this paper The results showthat our proposed algorithm can efficiently addressthe online SFC deployment problem
The remainder of this paper is organized as followsSection 2 provides an overview of related works Section 3provides the problem descriptions and formulation InSection 4 we describe our SFC deployment algorithm basedon the layered strategies of physical networks and eval-uation of physical network nodes Section 5 shows thesimulation results and analysis Section 6 summarizes thispaper
2 Related Work
NFV aims to satisfy client demands with minimal resourceconsumptions (eg bandwidth consumption and computingcore consumption) and high performance (eg low latencyand high throughput) [20] An increasing number of studieshave focused on the more efficient deployment of social IoToriented SFC requests in various scenarios
The research in [39] focused on the suitability of differentarchitectures of data center for a resilient SFC and theplacement of VNFs with high availability constraints In[40] the joint VNF placement and path selection problemwere studied The authors proposed a chain deployment
algorithm to balance the chain length and reuse factor itstarget was to serve as many demands as possible underlimited resources The research in [41] designed a heuristicalgorithm to address the NFV-RA problem in a coordinatedmanner by splitting the SFC request In [42] the authormade a BCMP mixed queuing network to replace the SFCmodel and proposed the convex optimization problem toshorten the acceptance interval of the service chain Thesimulation results shown the algorithm in [42] can effec-tively adapt the resource usage to the network dynamicsand serve more demands than other existing algorithmsThe authors of [43] studied SFC deployment in a multi-domain network and proposed a vertex-centric distributedcomputing algorithm to find all feasible SFC deploymentmethods of user requests in the distributed orchestrationframework then the algorithm selects the most appropriatescheme to deploy the SFC to achieve efficient performanceIn [44] the researcher discussed the phenomenon that theVNFs may sprawl across the network due to inefficientmapping during the SFC orchestration process and designedan efficient greedy heuristic algorithm to solve the problemThe author in [45] studied the service availability constraintsin a data center and designed a heuristic algorithm to deploythe SFC In the situation of distributed NFVs the authors in[46] studied the security level of a network security frame-work The authors proposed a new optimization algorithmto avoid a single point of failure in a bottleneck problemand obtained reasonable performance relative to that of acommon device Li et al [47] presented a real-time resource-distributing system for NFV which integrates timing analysiswith other algorithms such as timing abstraction SFC con-solidation and linear programming algorithm to efficientlydistribute network resources while considering latency con-straints
Many researchers have studied the deployment of thesocial IoT oriented VNF or SFC and a few of them haveattached importance to the total bandwidth consumptionYe et al [17] focused on the joint topology design and SFCdeployment problem to minimize the total consumption ofbandwidth resources Their main idea was to reduce band-width usage and cost by prioritizing VNF deployment In[48] the authors proposed and analyzed two deploymentstrategies HSD and VSD whose performance was analyzedin terms of cost They claimed that HSD had better per-formance in terms of an even distribution of the load overall servers access links and ToR switches The authors in[49] combined NFV with cloud computing and designeda bandwidth-guaranteed SFC-placing algorithm then theauthors tested the performance of their algorithm in a datacenter The simulation results showed that the heuristicalgorithm had excellent performance In [38] the authorsdesigned a forecast-assisted online SFC deployment algo-rithm that included the prediction of future VNF require-ments The simulation results showed that the algorithm in[38] could reduce the blocking probability of SFC requestsand effectively improve the profit of the service provider fromSFC deployment
Overall bandwidth is the most prominently expensiveresource that directly affects the number of demands that the
4 Security and Communication Networks
Client Client
Client
Client
90
f150
f270
f380 90
N4N5N3
N1
N2 N6
N7
N8
70
7080
8050
(a)
Client
Client
8070
70
N4N5N3
N1
N2 N6
N7
N8
Client Clientf150
f270
f380 90
(b)
Figure 2 Two examples of SFC deployment
network can satisfy online and the capacity of the physicalnetworkThus in this research we focused on the bandwidthconsumption and designed an efficient algorithm to provisionsocial IoT oriented SFC requests
3 Problem Formulation
In this paper we studied the online social IoT oriented SFCdeployment problem We considered a situation in whicheach SFC request has a pair of start and terminal nodesattached to the given physical network nodes and a specificsequence of VNFs that make up continuous functions Wehave to deploy these VNFs onto the corresponding physicalnodes and then organize the VNFs to form an SFC Toreduce the bandwidth resource consumption we need to usefewer nodes to shorten the length of the SFC as much aspossible
A user request can be denoted as 119878 = (119865S 119864S)where 119865S = 1198911 1198912 119891m is the set of VNFs and119864S = 1198901 1198902 119890q represents the virtual link after SFCdeployment The real physical underlying network can bemodeled as G = (N L) where G is an undirected weightedgraph 119873 = 11987311198732 119873y is the set of physical nodesand 119871 = 1198711 1198712 119871k is the set of real links inthe physical underlying network We use 119875 to denote thephysical path that holds the SFC request We define 119862119879119861 todenote the total bandwidth consumption which is definedin
119862119879119861 = sum119871 119894 119890119895isin119875 119878
119862119871 119894 119890119895119861 (1)
where119871 119894 119890119895 denotes that virtual link 119890119895 is deployed on the linkLi and 119862119871 119894 119890119895119861 is the bandwidth consumption of virtual link 119890119895which is deployed on the linkLi Wedefine119877119873119894119862 as the availablecomputing resources of the physical node119873119894 and 119862119873119894 119891119895119873 as thecomputing resource requirements of VNF 119891119895 which wouldbe deployed on the node Ni 119877119871 119894119861 is the available bandwidthresources of the physical link 119871119894
To deploy an SFC we must map the VNFs to some nodeand the virtual links 119864S to some real links in the physicalnetwork the path 119875 119878 which would hold the SFCs musthave sufficient nodes and computing resources to deploythe corresponding VNFs and the physical links must havesufficient available bandwidth for communication amongthose VNFs In addition the available bandwidth resourcesmust satisfy the requirements of the 119864S in the correspondinguser request Similar to [10 20] this paper assumes thateach VNF from the same SFC needs to use different physicalnetwork nodes Then the number of VNFs (denoted as Fs)should be less than the number of nodes of physical pathP 119878(denoted by 119875 119878) Then the problem of SFC deploymentcan be formulated as follows
min sum119871 119894 119890119895isin119875 119878
119862119871 119894 119890119895119861119904119905 forall119871 119894 119890119895 isin 119875 119878
119877119871 119894119861 minus 119862119871 119894 119890119895119861 ge 0forall119873119894 119891119895 isin 119875 119878119877119873119894119862 minus 119862119873119894 119891119895119873 ge 0119875 119878 minus 119865119904 ge 0
(2)
Formula (2) is used to model the SFC deployment prob-lem which can minimize the bandwidth consumptions whilefinishing SFC deployment There must be sufficient availablecomputing resources to deploy corresponding SFCs and thebandwidth must be sufficient to satisfy the communicationdemands among the corresponding VNFs In addition theremust be sufficient nodes in the physical path P to deploy theVNFs of the SFC request
Figure 2 shows two different examples of provisioningan SFC request both of which successfully deploy the SFCrequest and satisfy the clientsrsquo demands Nodes N1 and N8are the requested client node and destination client noderespectively There are three VNFs (ie f 1 f 2 and f 3) inthe SFC request Client node N1 needs to transmit 50 units
Security and Communication Networks 5
of traffic to VNF f 1 Between VNFs f 1 and f 2 70 unitsof information need to be transmitted Between VNFs f 2and f 3 80 units of traffic need to be transmitted Thereis a 90-unit transmission demand from VNF f 3 to thedestination client node N8 As shown in Figure 2(a) wedeploy the VNFs f 1 f 2 and f 3 onto the physical nodesN2 N5 and N6 respectively Then we find the path P =1198731-11987321198732-11987331198733-11987351198735-11987371198737-11987361198736-1198738 to hostthe SFC In this deployment scheme the total bandwidthconsumption is 440 units Although this approach success-fully deploys the SFC request it wastes bandwidth resourcesIn Figure 2(b) the VNFs f 1 f 2 and f 3 are deployed ontothe physical nodes N1 N6 and N8 respectively Then wefind a short path P = 1198731-11987321198732-11987361198736-1198738 (shown asthe red line in Figure 2(b)) which can also deploy all VNFsfrom the user requests However the total consumptionof bandwidth resources of this scheme is only 220 unitswhich is approximately half of the consumption of thedeployment in Figure 2(a) In Figure 2(b) path P is theshortest path to deploy this SFC request and this schemecan reduce more bandwidth consumption Mapping theSFC to the path in Figure 2(b) the service network candeploy more SFC with other links which is impossible inFigure 2(a)
Therefore different deployment schemes significantlyaffect the bandwidth consumptions and thus affect the avail-able network capacity An efficient algorithm is important andurgently needed to better deploy SFC requests and reducebandwidth consumption
4 Algorithm Design
Since the optimal SFC deployment problem is NP-hard [3537] we design an efficient algorithm for solving our researchproblem in this section Our proposed algorithm employsthe layered strategies of physical networks and evaluationof physical network nodes to optimize the consumption ofbandwidth resources which is denoted by OSFCD-LSEMThe basic idea of OSFCD-LSEM is to find the shortest pathsto deploy SFC requests while saving as many bandwidthresources as possible to satisfy more user requests Whena user demand arrives the OSFCD-LSEM algorithm beginsto handle it First calling Algorithm 2 the OSFCD-LSEMalgorithm layers the physical underlying network and obtainsthe information from the nodes and links in each layer ofthe physical network Then it calls Algorithm 3 to performan evaluation of nodes in the network and select some nodesthat are most suitable to host the VNFs of this SFC requestFinally it connects the deployed VNFs by using a shortestpath to fulfill an SFC By layered strategies and selection ofthe nodes for VNF deployment the OSFCD-LSEMalgorithmcan deploy VNFs in the most suitable nodes and deploythe SFC requests in appropriate simple paths to save morebandwidth resources The physical path must contain thestart node Nr and the terminal node Na Furthermore thenodes in the physical path must have sufficient resources tohost the VNFs of the user request
In the proposed OSFCD-LSEM algorithm we providesome definitions of variables 119866119871 is the physical network after
layering and 119881119883 is used to denote the set of nodes in the X-th layer The X-th layer is denoted as LX 119866119883119871 is the innerlayered network in the X-th layer (LX) and we used LY todenote the Y-th layer in 119866119883119871 119881119894(119883119884) is the set of nodes in theLY of 119866119883119871 which includes the node Ni 119864119883 is the set of linksthat connect the nodes from LX and LX-1 119864119894(119883119884) denotesthe corresponding links that connect the nodes in the LY-1about nodeNi and 119871119894119872119860119883 is the correspondingmaximal layerof node Ni LMAX is the layer number of GL 119873119879 is the nodenumber of the physical network and 119871119879 is the link numberof the physical network G The OSFCD-LSEM algorithm isshown in Algorithm 1
Next Algorithm 2 is responsible for handling hierarchicalphysical networks in our algorithm Algorithm 2 layers theentire network to obtain the layering information of thenodes and links in the network and outputs the results toother algorithms Therefore Algorithm 2 is the basis of ourSFC request deployment scheme and physical network nodeevaluation The physical network layering can be formulatedas follows
119866119871 =119871119872119860119883sum119883=1
(119881119883 119864119883) +119871119872119860119883sum119883=2
119866119883119871 (3)
119866119883119871 = sum119881119894isin119881119883
119871119894119872119860119883sum119884=1
(119881119894(119883119884) 119864119894(119883119884)) (4)
119871119872119860119883sum119883=1
119881119883 minus 119873119879 ge 0 (5)
119871119872119860119883sum119883=2
119864119883 +119871119872119860119883sum119883=2
sum119881119894isin119881119883
119871119894119872119860119883sum119884=2
119864119894(119883119884) minus 119871119879 = 0 (6)
In (3) two parts make up 119866119871 one part is the inner layernetwork 119866119883119871 about the X-th layer (LX) and the other partis the overall layered network The layering process beginsfrom the request node 119873119903 thus 1198811 = 119873119903 1198641 = 0 and1198661119871 = 0 Equation (4) shows that to make 119866119871 closer to thephysical network G each layer except layer L1 should obtainits inner layer node-related information The OSFCD-LSEMalgorithmcanobtain amore precise evaluation of the physicalnetwork and more efficiently deploy the corresponding SFCrequests In (5) the equation describes that all nodes in thelayered physical network must be in the corresponding layerIn (6) the equation shows that each link should be in thecorresponding layer or inner layer
In Figure 3 we give an example of layering a networkFigure 3(a) shows the topology of a physical network andit has a total of 10 nodes Figure 3(b) provides the detailedprocessing procedure for layering the network topology Weassume that the start node 119873119903 in the request is node N1 andthat the terminal node 119873119886 in the request is node N9 Firstour algorithm places the start node N1 into L1 (N1 is theonly node in V1 of L1) and places nodes N2 N3 and N4into L2 because they directly have links to node N1 Thenour algorithm places nodesN5N6 andN9 all of which havelinks to some nodes in the L2 (ie nodes N2 N3 and N4)
6 Security and Communication Networks
Input (1) SFC request (2) physical network 119866Output SFC deployment scheme(1) Receive a user request(2) Initialize Path = [](3) 119873119886 997888rarr Path119873119871 =119873119886(4) Layer the topology Algorithm 2(119873119903119873119871 119866)(5) Get 119871119860 the layers that destination client belongs to(6) while 119871119878 gt max(119871119860) + sum119871119872119860119883119883=1 max(119871119894119872119860119883) forall119873119894 isin 119881119883 do(7) if max119871119860 = 119871119872119860119883(8) 119873119879119864119872119875 = Algorithm 3( true max119871119860)(9) 119873119879119864119872119875 997888rarr Path(10) 119873119871 =119873119879119864119872119875(11) else(12) 119873119879119864119872119875 = Algorithm 3( false max119871119860)(13) 119873119879119864119872119875 997888rarr Path(14) 119873119871=119873119879119864119872119875(15) end if(16) 119871119878 = 119871119878 - 1(17) VNF 997888rarr 119873119879119864119872119875(18) Algorithm 2(119873119903119873119871 119866)(19) Update 119871119860(20) end while(21) if 119871119878 lemax(119871119860)(22) Select min LX isin 119871119860 ampamp LX gt 119871119878(23) while 119873119903 notin Path do(24) 119873119879119864119872119875 = Algorithm 3 ( true LX)(25) 119873119879119864119872119875 997888rarr Path(26) 119873119871 =119873119879119864119872119875(27) LX = LX - 1(28) VNF997888rarr 119873119879119864119872119875(29) end while(30) end if(31) SFC deployment scheme is Path
Algorithm 1 OSFCD-LSEM algorithm
into L3 (we specify that all nodes can only belong to onelayer except the terminal node N9 thus node N4 cannot bepart of L3 even though it connects with node N3 which isin L2)
In our layered network except for the destination clientnode N9 the nodes in one layer must have links to the nodesin the previous layer Thus nodes N7 N8 and N10 directlyhave links to the nodes in L3 (ie nodes N5 N6 and N9)whereas N10 connects only with destination client node N9among the three nodes in L3 NodeN10 should not be placedin layer L4 Thus we place only nodes N7 and N8 into L4Nodes N9 and N10 connect with nodes N7 and N8 in L4thus we place nodes N9 and N10 in layer L5 and becauseN9 connects with node N10 our algorithm places node N9in L6 When ten nodes in G belong to the correspondinglayers the overall network-layered processing finishes Thusthe OSFCD-LSEM algorithm can guarantee finding a pathwithout loops For each LX we also need to layer and obtainthe inner layer 119866119883119871 In the example shown by Figure 3(b) thelayer L2 has an inner layer 1198662119871 that includes two layers For1198662119871 in Figure 3(b) each layer X le LMAX and each node Ni isinVX should be set as the start node 119873119903 let 119873119886 = 0 Then we
obtain their inner layer information In 1198662119871 both 1198711198733119872119860119883 and1198711198734119872119860119883 are equal to 2 whereas 1198711198732119872119860119883 is equal to 1Overall the physical network is layered into six layers
with two inner layer networks associated with nodes N3 andN4The start node119873119903 is the only one in the first layer and theterminal node119873119886 is in three layers including L3 L5 and L6Thus we obtain three paths that can be used between119873119903 and119873119886 119871119875 is used to denote the length of path which is equal tothe number of VNFs that the path can hold LS is the lengthof an SFC which is equal to the VNF number of the SFCrequest
Algorithm 3 evaluates the nodes in the physical networkand selects a node which is most suitable to deploy therequired VNF After obtaining the layering informationAlgorithm 3makes the decision regarding whether themulti-ple links can satisfy the user request When themaximal layernumber in 119871119860 of the terminal nodes 119873119886 which is obtainedfrom the sumof all inner layers is smaller than the SFC length119871119878 in the user request the physical network cannot satisfy theuser request For example if we need to deploy an SFC intothe physical network in Figure 3(a) the start and terminalnodes are N1 and N9 The maximum layer number in 119871119860 is
Security and Communication Networks 7
Input (1) 119873119903 (2) 119873119886 (3) physical network GOutput 119866119871(1) 119873119903 997888rarr 1198811 119871119872119860119883 = L1(2) for 119881119871119872119860119883 = 0 119873119898 = 119873119886 do(3) for each119873119899 isin 119866 do(4) if 119873119898 larrrarr 119873119899 ampamp 119873119899 notin sum1198711198721198601198831 119881119883(5) 119873119899 997888rarr 119881119871119872119860119883+1 (6) else if 119873119898 larrrarr 119873119899 ampamp 119873119899 isin sum1198711198721198601198831 119881119883ampamp119873119899 =119873119886(7) 119873119899 997888rarr 119881119871119872119860119883+1 (8) end if(9) end for(10) 119871119872119860119883 ++(11) end for(12) for LX le 119871119872119860119883 do(13) for 119873119898 isin 119881119883 do(14) 119873119898 997888rarr 119881(1198831)(15) 119871119898119872119860119883 = 1198711198981(16) for 119881(119883119871119898119872119860119883) = 0 do(17) if 119873119899 isin 119881119883ampamp119873119898 larrrarr 119873119899ampamp119873119899 notin sum1198711198981198721198601198831 119866119883L(18) 119873119899 997888rarr 119881(119883119871119898119872119860119883)(19) end if(20) 119871119898119872119860119883 ++(21) end for(22) end for(23) end for(24) return 119866119871
Algorithm 2 Network layered processing
Input (1) SFC request(2) 119866119871(3) bool direction(4) X 119873119871 isin 119881119883Output the node119873119862 which has the minimum value of 120575(1) Temp = +infin(2) int i= 0(3) if direction is true(4) i= X-1(5) end if(6) if direction is false(7) i= X+1(8) end if(9) for 119873119898 isin 119881119894 do(10) if 119873119898 larrrarr 119873119871(11) if 119861119904119894 119898 gt 119861119903119894 119898ampamp119861119904119890 119898 gt 119861119903119890 119898 ampamp 119862119904 119898 gt 119862119903(12) Compute 120575 according to Equation (7)(13) if 120575 lt Temp(14) Temp = 120575(15) 119873119862 =119873119898(16) end if(17) end if(18) end if(19) end for(20) return 119873119862
Algorithm 3 Evaluation of related nodes
8 Security and Communication Networks
N1
N2
N3
N4 N6
N5
N8 N10
N9
N7
(a) Physical network
L1 V1 L3 V3L2 V2 L4 V4 L5 V5 L6 V6
N2
N3
N4
N5
N6
N9
N7
N8
N9
N10
N9N1
N4 N3N3 N4
6(21)3
6(22)3
6(21)4
6(22)4
(b) Processing procedure of layering
Figure 3 An example of a layered physical network
6 and there is an inner layer in layer L2The total sum is 7 sothis network topology can satisfy only an SFC request whoselength is no more than 7 An SFC request whose number ofVNFs ismore than 7 is too large for this network However aslong as the number of VNFs is not greater than the numberof nodes in the network our OSFCD-LSEM algorithm cantry to find more paths to deploy the longer SFC It may needadditional time and capacity of bandwidth resources becausethere are too many VNFs that need to be deployed Theproposed Algorithm 3 can search nodes in both positive andnegative directions When addressing an SFC that is longerthan the abovementioned sum Algorithm 3 commonly findsa suitable node in the layer119881119871119860+1 instead of in the upper layer119881119871119860minus1 then the algorithm can increase the length of the path(119871119875) However an extreme situation should be consideredsuch as when the node is in the last layer L119871119872119860119883 For thissituation our algorithm will find a node in the upper layer119881119871119872119860119883minus1 and run Algorithm 2 again
Finally we must evaluate the nodes from each layer ofthe layered network and select some nodes to map the VNFsAlgorithm 3 uses the abovementioned strategy to find a pathfrom 119873119886 to 119873119903 to host an SFC request and satisfy the userdemand Algorithm 3 selects the nodes from each layer using(7) and (8) The selected node must directly link to thenode in the next layer VN and must have sufficient resources
to deploy the VNFs and satisfy the corresponding functionrequirements
120575 = min(119861119904119894 119898 minus 119861119903119894 V119899119891119894119895119861119904119894 119898 119861119904119890 119898 minus 119861119903119890 V119899119891119894119895119861119904119890 119898 )
times 119862119904 119898 minus 119862119903 V119899119891119894119895119862119904 119898(7)
119862119904 119898 = 119862119905119900119905119886119897 119898 minus sum119878119894isin119878119865119862 119900119899119897119894119899119890
sumV119899119891119894119895isin119878119894
119862119903 V119899119891119894119895 times 120572V119899119891119894119895 119898 (8)
In (7) we use 120575 to denote the nodersquos appropriateness forthe VNF in the user request 119861119904119894 119898 is the idle bandwidth ofall links that connect node Nm with the nodes in the nextlayer 119881119873 and 119861119903119894 V119899119891119894119895 is the requested bandwidth resourcefrom vnf ij to the vnf ij+1 where V119899119891119894j is the j-th VNF in theSi (ie the i-th SFC request) 119861119904119890 119898 is the idle bandwidthof the path that connects node Nm with the nodes in theupper layer 119881119880 and 119861119903119890 V119899119891119894119895 is the requested bandwidthresource for making vnf ij connect with the last VNF 119862119904 119898represents the idle computing resources of node Nm and119862119903 V119899119891119894119895 represents the requested computing resources of thevnf ij In (8) Ctotal m represents the total computing resourcesof node Nm SFConline is the set of online SFC requests and
Security and Communication Networks 9
(a) US-Net topology (b) China-Net topology
Figure 4 Two real network topologies used in our simulations
120572V119899119891119894119895 119898 denotes whether vnf ij is deployed on node Nm Wecan evaluate whether a physical node is suitable to deploythe corresponding VNF according to the value of 120575 Afterphysical node evaluation we select the nodewith theminimalvalue of 120575 to hold the corresponding VNF
5 Simulation Results and Analysis
To evaluate the performance of our algorithm we compareour algorithm with two existing algorithms ie closed-loopwith critical mapping feedback (CCMF) [17] and key-VNFdeploy first (KVDF) [38] CCMF deploys the VNFs that havemore resource requirement priorities to optimize the totalconsumption of bandwidth resources KVDF focuses on therelation between different VNFs in the same SFC and therelation among different SFCs According to the relation andinfluence KVDF prioritizes the deployment of the key-VNFand key-SFC
To evaluate the performance of algorithms in differentnetwork scenarios we evaluate the performance of comparedalgorithms in moderate-scale and large-scale networks theUS-Network [50] (shown in Figure 4(a)) and the China-Network [51] (shown in Figure 4(b)) We use GT-ITM [52] togenerate the moderate-scale and large-scale network topolo-gies In moderate-scale network topologies there are approx-imately 100 nodes and 400 links In large-scale networktopologies there are approximately 300 nodes and 1000 linksTheUS-Net topology has 24 nodes and 43 links Additionallythe China-Net topology has 55 nodes and 103 links In thesenetwork topologies the bandwidth resource of all links isuniformly distributed at 100sim200 units and the computingresource of all nodes is set as 10 units
In our simulations we set the following parametersaccording to the existing work [49] The computer randomlygenerates user requests with lengths (ie Ls) from 5 to 14 andthese online SFC requests arrive as a Poisson process For eachphysical network and for a given 119871 119904 we randomly generate10000 SFC requests with the request client and destinationclient nodes which are randomly assigned to the physicalnodesThe computing resource demand of each VNF followsa uniform distribution U (1 3) and the bandwidth resource
demand of each virtual link follows a uniform distribution U(20 50)
With the increasing number of users and SFC requeststhe deployment of SFC in a static network becomes increas-ingly challenging Thus the network scalability must beimproved For a network-aware scaling strategy it is betterto extend the network instead of changing the network Innetwork G we define the evaluation of information 119866119878 in
119866119878 = 119871119872119860119883sum119883=1
sum119873119894isin119881119883
(119862119904 + 119861119904119894 + 119861119904119890) (9)
The OSFCD-LSEM algorithm layers the physical net-work obtains the layer which has minimum resourcesanalyzes its inner layer information and obtains the ldquoweakrdquonodes or links that influence the capacity of the networkThen the OSFCD-LSEM algorithm extends their resourcesto secure a more robust network
We obtained the simulation results by averaging theresults of multiple simulations We used an Ubuntu virtualmachine running on a computer with a 37 GHz Intel Corei3-4170 and 4 GB of RAM to run the simulations And thealgorithms were coded in Java programming language
From Figure 5 we can see the simulation results ofour OSFCD-LSEM algorithm in a moderate-scale networkFigure 5(a) shows the information of the physical networkWe can see that L8 limits the network capacity and will affectthe deployment of SFC Figure 5(b) shows the information ofthe physical nodes in L8 In L8 node-67 has the minimalamount of bandwidth and node-72 has the minimal amountof computing resources Both node-67 and node-72 are theldquoweak pointsrdquo in the physical network If we can increasetheir corresponding resources the capacity of the physicalnetwork will be enhanced For network operators increasingthe corresponding resources to the corresponding nodes andlinks is necessary and highly beneficial moreover they canalso obtain a more robust network
Figure 6 shows the simulation results on the US-Netnetwork Figure 6(a) shows the information about each layerin the overall network and we find that L4 may be theldquoweak pointrdquo of the physical network because of the lackof computing and bandwidth resources Figure 6(b) shows
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
4 Security and Communication Networks
Client Client
Client
Client
90
f150
f270
f380 90
N4N5N3
N1
N2 N6
N7
N8
70
7080
8050
(a)
Client
Client
8070
70
N4N5N3
N1
N2 N6
N7
N8
Client Clientf150
f270
f380 90
(b)
Figure 2 Two examples of SFC deployment
network can satisfy online and the capacity of the physicalnetworkThus in this research we focused on the bandwidthconsumption and designed an efficient algorithm to provisionsocial IoT oriented SFC requests
3 Problem Formulation
In this paper we studied the online social IoT oriented SFCdeployment problem We considered a situation in whicheach SFC request has a pair of start and terminal nodesattached to the given physical network nodes and a specificsequence of VNFs that make up continuous functions Wehave to deploy these VNFs onto the corresponding physicalnodes and then organize the VNFs to form an SFC Toreduce the bandwidth resource consumption we need to usefewer nodes to shorten the length of the SFC as much aspossible
A user request can be denoted as 119878 = (119865S 119864S)where 119865S = 1198911 1198912 119891m is the set of VNFs and119864S = 1198901 1198902 119890q represents the virtual link after SFCdeployment The real physical underlying network can bemodeled as G = (N L) where G is an undirected weightedgraph 119873 = 11987311198732 119873y is the set of physical nodesand 119871 = 1198711 1198712 119871k is the set of real links inthe physical underlying network We use 119875 to denote thephysical path that holds the SFC request We define 119862119879119861 todenote the total bandwidth consumption which is definedin
119862119879119861 = sum119871 119894 119890119895isin119875 119878
119862119871 119894 119890119895119861 (1)
where119871 119894 119890119895 denotes that virtual link 119890119895 is deployed on the linkLi and 119862119871 119894 119890119895119861 is the bandwidth consumption of virtual link 119890119895which is deployed on the linkLi Wedefine119877119873119894119862 as the availablecomputing resources of the physical node119873119894 and 119862119873119894 119891119895119873 as thecomputing resource requirements of VNF 119891119895 which wouldbe deployed on the node Ni 119877119871 119894119861 is the available bandwidthresources of the physical link 119871119894
To deploy an SFC we must map the VNFs to some nodeand the virtual links 119864S to some real links in the physicalnetwork the path 119875 119878 which would hold the SFCs musthave sufficient nodes and computing resources to deploythe corresponding VNFs and the physical links must havesufficient available bandwidth for communication amongthose VNFs In addition the available bandwidth resourcesmust satisfy the requirements of the 119864S in the correspondinguser request Similar to [10 20] this paper assumes thateach VNF from the same SFC needs to use different physicalnetwork nodes Then the number of VNFs (denoted as Fs)should be less than the number of nodes of physical pathP 119878(denoted by 119875 119878) Then the problem of SFC deploymentcan be formulated as follows
min sum119871 119894 119890119895isin119875 119878
119862119871 119894 119890119895119861119904119905 forall119871 119894 119890119895 isin 119875 119878
119877119871 119894119861 minus 119862119871 119894 119890119895119861 ge 0forall119873119894 119891119895 isin 119875 119878119877119873119894119862 minus 119862119873119894 119891119895119873 ge 0119875 119878 minus 119865119904 ge 0
(2)
Formula (2) is used to model the SFC deployment prob-lem which can minimize the bandwidth consumptions whilefinishing SFC deployment There must be sufficient availablecomputing resources to deploy corresponding SFCs and thebandwidth must be sufficient to satisfy the communicationdemands among the corresponding VNFs In addition theremust be sufficient nodes in the physical path P to deploy theVNFs of the SFC request
Figure 2 shows two different examples of provisioningan SFC request both of which successfully deploy the SFCrequest and satisfy the clientsrsquo demands Nodes N1 and N8are the requested client node and destination client noderespectively There are three VNFs (ie f 1 f 2 and f 3) inthe SFC request Client node N1 needs to transmit 50 units
Security and Communication Networks 5
of traffic to VNF f 1 Between VNFs f 1 and f 2 70 unitsof information need to be transmitted Between VNFs f 2and f 3 80 units of traffic need to be transmitted Thereis a 90-unit transmission demand from VNF f 3 to thedestination client node N8 As shown in Figure 2(a) wedeploy the VNFs f 1 f 2 and f 3 onto the physical nodesN2 N5 and N6 respectively Then we find the path P =1198731-11987321198732-11987331198733-11987351198735-11987371198737-11987361198736-1198738 to hostthe SFC In this deployment scheme the total bandwidthconsumption is 440 units Although this approach success-fully deploys the SFC request it wastes bandwidth resourcesIn Figure 2(b) the VNFs f 1 f 2 and f 3 are deployed ontothe physical nodes N1 N6 and N8 respectively Then wefind a short path P = 1198731-11987321198732-11987361198736-1198738 (shown asthe red line in Figure 2(b)) which can also deploy all VNFsfrom the user requests However the total consumptionof bandwidth resources of this scheme is only 220 unitswhich is approximately half of the consumption of thedeployment in Figure 2(a) In Figure 2(b) path P is theshortest path to deploy this SFC request and this schemecan reduce more bandwidth consumption Mapping theSFC to the path in Figure 2(b) the service network candeploy more SFC with other links which is impossible inFigure 2(a)
Therefore different deployment schemes significantlyaffect the bandwidth consumptions and thus affect the avail-able network capacity An efficient algorithm is important andurgently needed to better deploy SFC requests and reducebandwidth consumption
4 Algorithm Design
Since the optimal SFC deployment problem is NP-hard [3537] we design an efficient algorithm for solving our researchproblem in this section Our proposed algorithm employsthe layered strategies of physical networks and evaluationof physical network nodes to optimize the consumption ofbandwidth resources which is denoted by OSFCD-LSEMThe basic idea of OSFCD-LSEM is to find the shortest pathsto deploy SFC requests while saving as many bandwidthresources as possible to satisfy more user requests Whena user demand arrives the OSFCD-LSEM algorithm beginsto handle it First calling Algorithm 2 the OSFCD-LSEMalgorithm layers the physical underlying network and obtainsthe information from the nodes and links in each layer ofthe physical network Then it calls Algorithm 3 to performan evaluation of nodes in the network and select some nodesthat are most suitable to host the VNFs of this SFC requestFinally it connects the deployed VNFs by using a shortestpath to fulfill an SFC By layered strategies and selection ofthe nodes for VNF deployment the OSFCD-LSEMalgorithmcan deploy VNFs in the most suitable nodes and deploythe SFC requests in appropriate simple paths to save morebandwidth resources The physical path must contain thestart node Nr and the terminal node Na Furthermore thenodes in the physical path must have sufficient resources tohost the VNFs of the user request
In the proposed OSFCD-LSEM algorithm we providesome definitions of variables 119866119871 is the physical network after
layering and 119881119883 is used to denote the set of nodes in the X-th layer The X-th layer is denoted as LX 119866119883119871 is the innerlayered network in the X-th layer (LX) and we used LY todenote the Y-th layer in 119866119883119871 119881119894(119883119884) is the set of nodes in theLY of 119866119883119871 which includes the node Ni 119864119883 is the set of linksthat connect the nodes from LX and LX-1 119864119894(119883119884) denotesthe corresponding links that connect the nodes in the LY-1about nodeNi and 119871119894119872119860119883 is the correspondingmaximal layerof node Ni LMAX is the layer number of GL 119873119879 is the nodenumber of the physical network and 119871119879 is the link numberof the physical network G The OSFCD-LSEM algorithm isshown in Algorithm 1
Next Algorithm 2 is responsible for handling hierarchicalphysical networks in our algorithm Algorithm 2 layers theentire network to obtain the layering information of thenodes and links in the network and outputs the results toother algorithms Therefore Algorithm 2 is the basis of ourSFC request deployment scheme and physical network nodeevaluation The physical network layering can be formulatedas follows
119866119871 =119871119872119860119883sum119883=1
(119881119883 119864119883) +119871119872119860119883sum119883=2
119866119883119871 (3)
119866119883119871 = sum119881119894isin119881119883
119871119894119872119860119883sum119884=1
(119881119894(119883119884) 119864119894(119883119884)) (4)
119871119872119860119883sum119883=1
119881119883 minus 119873119879 ge 0 (5)
119871119872119860119883sum119883=2
119864119883 +119871119872119860119883sum119883=2
sum119881119894isin119881119883
119871119894119872119860119883sum119884=2
119864119894(119883119884) minus 119871119879 = 0 (6)
In (3) two parts make up 119866119871 one part is the inner layernetwork 119866119883119871 about the X-th layer (LX) and the other partis the overall layered network The layering process beginsfrom the request node 119873119903 thus 1198811 = 119873119903 1198641 = 0 and1198661119871 = 0 Equation (4) shows that to make 119866119871 closer to thephysical network G each layer except layer L1 should obtainits inner layer node-related information The OSFCD-LSEMalgorithmcanobtain amore precise evaluation of the physicalnetwork and more efficiently deploy the corresponding SFCrequests In (5) the equation describes that all nodes in thelayered physical network must be in the corresponding layerIn (6) the equation shows that each link should be in thecorresponding layer or inner layer
In Figure 3 we give an example of layering a networkFigure 3(a) shows the topology of a physical network andit has a total of 10 nodes Figure 3(b) provides the detailedprocessing procedure for layering the network topology Weassume that the start node 119873119903 in the request is node N1 andthat the terminal node 119873119886 in the request is node N9 Firstour algorithm places the start node N1 into L1 (N1 is theonly node in V1 of L1) and places nodes N2 N3 and N4into L2 because they directly have links to node N1 Thenour algorithm places nodesN5N6 andN9 all of which havelinks to some nodes in the L2 (ie nodes N2 N3 and N4)
6 Security and Communication Networks
Input (1) SFC request (2) physical network 119866Output SFC deployment scheme(1) Receive a user request(2) Initialize Path = [](3) 119873119886 997888rarr Path119873119871 =119873119886(4) Layer the topology Algorithm 2(119873119903119873119871 119866)(5) Get 119871119860 the layers that destination client belongs to(6) while 119871119878 gt max(119871119860) + sum119871119872119860119883119883=1 max(119871119894119872119860119883) forall119873119894 isin 119881119883 do(7) if max119871119860 = 119871119872119860119883(8) 119873119879119864119872119875 = Algorithm 3( true max119871119860)(9) 119873119879119864119872119875 997888rarr Path(10) 119873119871 =119873119879119864119872119875(11) else(12) 119873119879119864119872119875 = Algorithm 3( false max119871119860)(13) 119873119879119864119872119875 997888rarr Path(14) 119873119871=119873119879119864119872119875(15) end if(16) 119871119878 = 119871119878 - 1(17) VNF 997888rarr 119873119879119864119872119875(18) Algorithm 2(119873119903119873119871 119866)(19) Update 119871119860(20) end while(21) if 119871119878 lemax(119871119860)(22) Select min LX isin 119871119860 ampamp LX gt 119871119878(23) while 119873119903 notin Path do(24) 119873119879119864119872119875 = Algorithm 3 ( true LX)(25) 119873119879119864119872119875 997888rarr Path(26) 119873119871 =119873119879119864119872119875(27) LX = LX - 1(28) VNF997888rarr 119873119879119864119872119875(29) end while(30) end if(31) SFC deployment scheme is Path
Algorithm 1 OSFCD-LSEM algorithm
into L3 (we specify that all nodes can only belong to onelayer except the terminal node N9 thus node N4 cannot bepart of L3 even though it connects with node N3 which isin L2)
In our layered network except for the destination clientnode N9 the nodes in one layer must have links to the nodesin the previous layer Thus nodes N7 N8 and N10 directlyhave links to the nodes in L3 (ie nodes N5 N6 and N9)whereas N10 connects only with destination client node N9among the three nodes in L3 NodeN10 should not be placedin layer L4 Thus we place only nodes N7 and N8 into L4Nodes N9 and N10 connect with nodes N7 and N8 in L4thus we place nodes N9 and N10 in layer L5 and becauseN9 connects with node N10 our algorithm places node N9in L6 When ten nodes in G belong to the correspondinglayers the overall network-layered processing finishes Thusthe OSFCD-LSEM algorithm can guarantee finding a pathwithout loops For each LX we also need to layer and obtainthe inner layer 119866119883119871 In the example shown by Figure 3(b) thelayer L2 has an inner layer 1198662119871 that includes two layers For1198662119871 in Figure 3(b) each layer X le LMAX and each node Ni isinVX should be set as the start node 119873119903 let 119873119886 = 0 Then we
obtain their inner layer information In 1198662119871 both 1198711198733119872119860119883 and1198711198734119872119860119883 are equal to 2 whereas 1198711198732119872119860119883 is equal to 1Overall the physical network is layered into six layers
with two inner layer networks associated with nodes N3 andN4The start node119873119903 is the only one in the first layer and theterminal node119873119886 is in three layers including L3 L5 and L6Thus we obtain three paths that can be used between119873119903 and119873119886 119871119875 is used to denote the length of path which is equal tothe number of VNFs that the path can hold LS is the lengthof an SFC which is equal to the VNF number of the SFCrequest
Algorithm 3 evaluates the nodes in the physical networkand selects a node which is most suitable to deploy therequired VNF After obtaining the layering informationAlgorithm 3makes the decision regarding whether themulti-ple links can satisfy the user request When themaximal layernumber in 119871119860 of the terminal nodes 119873119886 which is obtainedfrom the sumof all inner layers is smaller than the SFC length119871119878 in the user request the physical network cannot satisfy theuser request For example if we need to deploy an SFC intothe physical network in Figure 3(a) the start and terminalnodes are N1 and N9 The maximum layer number in 119871119860 is
Security and Communication Networks 7
Input (1) 119873119903 (2) 119873119886 (3) physical network GOutput 119866119871(1) 119873119903 997888rarr 1198811 119871119872119860119883 = L1(2) for 119881119871119872119860119883 = 0 119873119898 = 119873119886 do(3) for each119873119899 isin 119866 do(4) if 119873119898 larrrarr 119873119899 ampamp 119873119899 notin sum1198711198721198601198831 119881119883(5) 119873119899 997888rarr 119881119871119872119860119883+1 (6) else if 119873119898 larrrarr 119873119899 ampamp 119873119899 isin sum1198711198721198601198831 119881119883ampamp119873119899 =119873119886(7) 119873119899 997888rarr 119881119871119872119860119883+1 (8) end if(9) end for(10) 119871119872119860119883 ++(11) end for(12) for LX le 119871119872119860119883 do(13) for 119873119898 isin 119881119883 do(14) 119873119898 997888rarr 119881(1198831)(15) 119871119898119872119860119883 = 1198711198981(16) for 119881(119883119871119898119872119860119883) = 0 do(17) if 119873119899 isin 119881119883ampamp119873119898 larrrarr 119873119899ampamp119873119899 notin sum1198711198981198721198601198831 119866119883L(18) 119873119899 997888rarr 119881(119883119871119898119872119860119883)(19) end if(20) 119871119898119872119860119883 ++(21) end for(22) end for(23) end for(24) return 119866119871
Algorithm 2 Network layered processing
Input (1) SFC request(2) 119866119871(3) bool direction(4) X 119873119871 isin 119881119883Output the node119873119862 which has the minimum value of 120575(1) Temp = +infin(2) int i= 0(3) if direction is true(4) i= X-1(5) end if(6) if direction is false(7) i= X+1(8) end if(9) for 119873119898 isin 119881119894 do(10) if 119873119898 larrrarr 119873119871(11) if 119861119904119894 119898 gt 119861119903119894 119898ampamp119861119904119890 119898 gt 119861119903119890 119898 ampamp 119862119904 119898 gt 119862119903(12) Compute 120575 according to Equation (7)(13) if 120575 lt Temp(14) Temp = 120575(15) 119873119862 =119873119898(16) end if(17) end if(18) end if(19) end for(20) return 119873119862
Algorithm 3 Evaluation of related nodes
8 Security and Communication Networks
N1
N2
N3
N4 N6
N5
N8 N10
N9
N7
(a) Physical network
L1 V1 L3 V3L2 V2 L4 V4 L5 V5 L6 V6
N2
N3
N4
N5
N6
N9
N7
N8
N9
N10
N9N1
N4 N3N3 N4
6(21)3
6(22)3
6(21)4
6(22)4
(b) Processing procedure of layering
Figure 3 An example of a layered physical network
6 and there is an inner layer in layer L2The total sum is 7 sothis network topology can satisfy only an SFC request whoselength is no more than 7 An SFC request whose number ofVNFs ismore than 7 is too large for this network However aslong as the number of VNFs is not greater than the numberof nodes in the network our OSFCD-LSEM algorithm cantry to find more paths to deploy the longer SFC It may needadditional time and capacity of bandwidth resources becausethere are too many VNFs that need to be deployed Theproposed Algorithm 3 can search nodes in both positive andnegative directions When addressing an SFC that is longerthan the abovementioned sum Algorithm 3 commonly findsa suitable node in the layer119881119871119860+1 instead of in the upper layer119881119871119860minus1 then the algorithm can increase the length of the path(119871119875) However an extreme situation should be consideredsuch as when the node is in the last layer L119871119872119860119883 For thissituation our algorithm will find a node in the upper layer119881119871119872119860119883minus1 and run Algorithm 2 again
Finally we must evaluate the nodes from each layer ofthe layered network and select some nodes to map the VNFsAlgorithm 3 uses the abovementioned strategy to find a pathfrom 119873119886 to 119873119903 to host an SFC request and satisfy the userdemand Algorithm 3 selects the nodes from each layer using(7) and (8) The selected node must directly link to thenode in the next layer VN and must have sufficient resources
to deploy the VNFs and satisfy the corresponding functionrequirements
120575 = min(119861119904119894 119898 minus 119861119903119894 V119899119891119894119895119861119904119894 119898 119861119904119890 119898 minus 119861119903119890 V119899119891119894119895119861119904119890 119898 )
times 119862119904 119898 minus 119862119903 V119899119891119894119895119862119904 119898(7)
119862119904 119898 = 119862119905119900119905119886119897 119898 minus sum119878119894isin119878119865119862 119900119899119897119894119899119890
sumV119899119891119894119895isin119878119894
119862119903 V119899119891119894119895 times 120572V119899119891119894119895 119898 (8)
In (7) we use 120575 to denote the nodersquos appropriateness forthe VNF in the user request 119861119904119894 119898 is the idle bandwidth ofall links that connect node Nm with the nodes in the nextlayer 119881119873 and 119861119903119894 V119899119891119894119895 is the requested bandwidth resourcefrom vnf ij to the vnf ij+1 where V119899119891119894j is the j-th VNF in theSi (ie the i-th SFC request) 119861119904119890 119898 is the idle bandwidthof the path that connects node Nm with the nodes in theupper layer 119881119880 and 119861119903119890 V119899119891119894119895 is the requested bandwidthresource for making vnf ij connect with the last VNF 119862119904 119898represents the idle computing resources of node Nm and119862119903 V119899119891119894119895 represents the requested computing resources of thevnf ij In (8) Ctotal m represents the total computing resourcesof node Nm SFConline is the set of online SFC requests and
Security and Communication Networks 9
(a) US-Net topology (b) China-Net topology
Figure 4 Two real network topologies used in our simulations
120572V119899119891119894119895 119898 denotes whether vnf ij is deployed on node Nm Wecan evaluate whether a physical node is suitable to deploythe corresponding VNF according to the value of 120575 Afterphysical node evaluation we select the nodewith theminimalvalue of 120575 to hold the corresponding VNF
5 Simulation Results and Analysis
To evaluate the performance of our algorithm we compareour algorithm with two existing algorithms ie closed-loopwith critical mapping feedback (CCMF) [17] and key-VNFdeploy first (KVDF) [38] CCMF deploys the VNFs that havemore resource requirement priorities to optimize the totalconsumption of bandwidth resources KVDF focuses on therelation between different VNFs in the same SFC and therelation among different SFCs According to the relation andinfluence KVDF prioritizes the deployment of the key-VNFand key-SFC
To evaluate the performance of algorithms in differentnetwork scenarios we evaluate the performance of comparedalgorithms in moderate-scale and large-scale networks theUS-Network [50] (shown in Figure 4(a)) and the China-Network [51] (shown in Figure 4(b)) We use GT-ITM [52] togenerate the moderate-scale and large-scale network topolo-gies In moderate-scale network topologies there are approx-imately 100 nodes and 400 links In large-scale networktopologies there are approximately 300 nodes and 1000 linksTheUS-Net topology has 24 nodes and 43 links Additionallythe China-Net topology has 55 nodes and 103 links In thesenetwork topologies the bandwidth resource of all links isuniformly distributed at 100sim200 units and the computingresource of all nodes is set as 10 units
In our simulations we set the following parametersaccording to the existing work [49] The computer randomlygenerates user requests with lengths (ie Ls) from 5 to 14 andthese online SFC requests arrive as a Poisson process For eachphysical network and for a given 119871 119904 we randomly generate10000 SFC requests with the request client and destinationclient nodes which are randomly assigned to the physicalnodesThe computing resource demand of each VNF followsa uniform distribution U (1 3) and the bandwidth resource
demand of each virtual link follows a uniform distribution U(20 50)
With the increasing number of users and SFC requeststhe deployment of SFC in a static network becomes increas-ingly challenging Thus the network scalability must beimproved For a network-aware scaling strategy it is betterto extend the network instead of changing the network Innetwork G we define the evaluation of information 119866119878 in
119866119878 = 119871119872119860119883sum119883=1
sum119873119894isin119881119883
(119862119904 + 119861119904119894 + 119861119904119890) (9)
The OSFCD-LSEM algorithm layers the physical net-work obtains the layer which has minimum resourcesanalyzes its inner layer information and obtains the ldquoweakrdquonodes or links that influence the capacity of the networkThen the OSFCD-LSEM algorithm extends their resourcesto secure a more robust network
We obtained the simulation results by averaging theresults of multiple simulations We used an Ubuntu virtualmachine running on a computer with a 37 GHz Intel Corei3-4170 and 4 GB of RAM to run the simulations And thealgorithms were coded in Java programming language
From Figure 5 we can see the simulation results ofour OSFCD-LSEM algorithm in a moderate-scale networkFigure 5(a) shows the information of the physical networkWe can see that L8 limits the network capacity and will affectthe deployment of SFC Figure 5(b) shows the information ofthe physical nodes in L8 In L8 node-67 has the minimalamount of bandwidth and node-72 has the minimal amountof computing resources Both node-67 and node-72 are theldquoweak pointsrdquo in the physical network If we can increasetheir corresponding resources the capacity of the physicalnetwork will be enhanced For network operators increasingthe corresponding resources to the corresponding nodes andlinks is necessary and highly beneficial moreover they canalso obtain a more robust network
Figure 6 shows the simulation results on the US-Netnetwork Figure 6(a) shows the information about each layerin the overall network and we find that L4 may be theldquoweak pointrdquo of the physical network because of the lackof computing and bandwidth resources Figure 6(b) shows
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 5
of traffic to VNF f 1 Between VNFs f 1 and f 2 70 unitsof information need to be transmitted Between VNFs f 2and f 3 80 units of traffic need to be transmitted Thereis a 90-unit transmission demand from VNF f 3 to thedestination client node N8 As shown in Figure 2(a) wedeploy the VNFs f 1 f 2 and f 3 onto the physical nodesN2 N5 and N6 respectively Then we find the path P =1198731-11987321198732-11987331198733-11987351198735-11987371198737-11987361198736-1198738 to hostthe SFC In this deployment scheme the total bandwidthconsumption is 440 units Although this approach success-fully deploys the SFC request it wastes bandwidth resourcesIn Figure 2(b) the VNFs f 1 f 2 and f 3 are deployed ontothe physical nodes N1 N6 and N8 respectively Then wefind a short path P = 1198731-11987321198732-11987361198736-1198738 (shown asthe red line in Figure 2(b)) which can also deploy all VNFsfrom the user requests However the total consumptionof bandwidth resources of this scheme is only 220 unitswhich is approximately half of the consumption of thedeployment in Figure 2(a) In Figure 2(b) path P is theshortest path to deploy this SFC request and this schemecan reduce more bandwidth consumption Mapping theSFC to the path in Figure 2(b) the service network candeploy more SFC with other links which is impossible inFigure 2(a)
Therefore different deployment schemes significantlyaffect the bandwidth consumptions and thus affect the avail-able network capacity An efficient algorithm is important andurgently needed to better deploy SFC requests and reducebandwidth consumption
4 Algorithm Design
Since the optimal SFC deployment problem is NP-hard [3537] we design an efficient algorithm for solving our researchproblem in this section Our proposed algorithm employsthe layered strategies of physical networks and evaluationof physical network nodes to optimize the consumption ofbandwidth resources which is denoted by OSFCD-LSEMThe basic idea of OSFCD-LSEM is to find the shortest pathsto deploy SFC requests while saving as many bandwidthresources as possible to satisfy more user requests Whena user demand arrives the OSFCD-LSEM algorithm beginsto handle it First calling Algorithm 2 the OSFCD-LSEMalgorithm layers the physical underlying network and obtainsthe information from the nodes and links in each layer ofthe physical network Then it calls Algorithm 3 to performan evaluation of nodes in the network and select some nodesthat are most suitable to host the VNFs of this SFC requestFinally it connects the deployed VNFs by using a shortestpath to fulfill an SFC By layered strategies and selection ofthe nodes for VNF deployment the OSFCD-LSEMalgorithmcan deploy VNFs in the most suitable nodes and deploythe SFC requests in appropriate simple paths to save morebandwidth resources The physical path must contain thestart node Nr and the terminal node Na Furthermore thenodes in the physical path must have sufficient resources tohost the VNFs of the user request
In the proposed OSFCD-LSEM algorithm we providesome definitions of variables 119866119871 is the physical network after
layering and 119881119883 is used to denote the set of nodes in the X-th layer The X-th layer is denoted as LX 119866119883119871 is the innerlayered network in the X-th layer (LX) and we used LY todenote the Y-th layer in 119866119883119871 119881119894(119883119884) is the set of nodes in theLY of 119866119883119871 which includes the node Ni 119864119883 is the set of linksthat connect the nodes from LX and LX-1 119864119894(119883119884) denotesthe corresponding links that connect the nodes in the LY-1about nodeNi and 119871119894119872119860119883 is the correspondingmaximal layerof node Ni LMAX is the layer number of GL 119873119879 is the nodenumber of the physical network and 119871119879 is the link numberof the physical network G The OSFCD-LSEM algorithm isshown in Algorithm 1
Next Algorithm 2 is responsible for handling hierarchicalphysical networks in our algorithm Algorithm 2 layers theentire network to obtain the layering information of thenodes and links in the network and outputs the results toother algorithms Therefore Algorithm 2 is the basis of ourSFC request deployment scheme and physical network nodeevaluation The physical network layering can be formulatedas follows
119866119871 =119871119872119860119883sum119883=1
(119881119883 119864119883) +119871119872119860119883sum119883=2
119866119883119871 (3)
119866119883119871 = sum119881119894isin119881119883
119871119894119872119860119883sum119884=1
(119881119894(119883119884) 119864119894(119883119884)) (4)
119871119872119860119883sum119883=1
119881119883 minus 119873119879 ge 0 (5)
119871119872119860119883sum119883=2
119864119883 +119871119872119860119883sum119883=2
sum119881119894isin119881119883
119871119894119872119860119883sum119884=2
119864119894(119883119884) minus 119871119879 = 0 (6)
In (3) two parts make up 119866119871 one part is the inner layernetwork 119866119883119871 about the X-th layer (LX) and the other partis the overall layered network The layering process beginsfrom the request node 119873119903 thus 1198811 = 119873119903 1198641 = 0 and1198661119871 = 0 Equation (4) shows that to make 119866119871 closer to thephysical network G each layer except layer L1 should obtainits inner layer node-related information The OSFCD-LSEMalgorithmcanobtain amore precise evaluation of the physicalnetwork and more efficiently deploy the corresponding SFCrequests In (5) the equation describes that all nodes in thelayered physical network must be in the corresponding layerIn (6) the equation shows that each link should be in thecorresponding layer or inner layer
In Figure 3 we give an example of layering a networkFigure 3(a) shows the topology of a physical network andit has a total of 10 nodes Figure 3(b) provides the detailedprocessing procedure for layering the network topology Weassume that the start node 119873119903 in the request is node N1 andthat the terminal node 119873119886 in the request is node N9 Firstour algorithm places the start node N1 into L1 (N1 is theonly node in V1 of L1) and places nodes N2 N3 and N4into L2 because they directly have links to node N1 Thenour algorithm places nodesN5N6 andN9 all of which havelinks to some nodes in the L2 (ie nodes N2 N3 and N4)
6 Security and Communication Networks
Input (1) SFC request (2) physical network 119866Output SFC deployment scheme(1) Receive a user request(2) Initialize Path = [](3) 119873119886 997888rarr Path119873119871 =119873119886(4) Layer the topology Algorithm 2(119873119903119873119871 119866)(5) Get 119871119860 the layers that destination client belongs to(6) while 119871119878 gt max(119871119860) + sum119871119872119860119883119883=1 max(119871119894119872119860119883) forall119873119894 isin 119881119883 do(7) if max119871119860 = 119871119872119860119883(8) 119873119879119864119872119875 = Algorithm 3( true max119871119860)(9) 119873119879119864119872119875 997888rarr Path(10) 119873119871 =119873119879119864119872119875(11) else(12) 119873119879119864119872119875 = Algorithm 3( false max119871119860)(13) 119873119879119864119872119875 997888rarr Path(14) 119873119871=119873119879119864119872119875(15) end if(16) 119871119878 = 119871119878 - 1(17) VNF 997888rarr 119873119879119864119872119875(18) Algorithm 2(119873119903119873119871 119866)(19) Update 119871119860(20) end while(21) if 119871119878 lemax(119871119860)(22) Select min LX isin 119871119860 ampamp LX gt 119871119878(23) while 119873119903 notin Path do(24) 119873119879119864119872119875 = Algorithm 3 ( true LX)(25) 119873119879119864119872119875 997888rarr Path(26) 119873119871 =119873119879119864119872119875(27) LX = LX - 1(28) VNF997888rarr 119873119879119864119872119875(29) end while(30) end if(31) SFC deployment scheme is Path
Algorithm 1 OSFCD-LSEM algorithm
into L3 (we specify that all nodes can only belong to onelayer except the terminal node N9 thus node N4 cannot bepart of L3 even though it connects with node N3 which isin L2)
In our layered network except for the destination clientnode N9 the nodes in one layer must have links to the nodesin the previous layer Thus nodes N7 N8 and N10 directlyhave links to the nodes in L3 (ie nodes N5 N6 and N9)whereas N10 connects only with destination client node N9among the three nodes in L3 NodeN10 should not be placedin layer L4 Thus we place only nodes N7 and N8 into L4Nodes N9 and N10 connect with nodes N7 and N8 in L4thus we place nodes N9 and N10 in layer L5 and becauseN9 connects with node N10 our algorithm places node N9in L6 When ten nodes in G belong to the correspondinglayers the overall network-layered processing finishes Thusthe OSFCD-LSEM algorithm can guarantee finding a pathwithout loops For each LX we also need to layer and obtainthe inner layer 119866119883119871 In the example shown by Figure 3(b) thelayer L2 has an inner layer 1198662119871 that includes two layers For1198662119871 in Figure 3(b) each layer X le LMAX and each node Ni isinVX should be set as the start node 119873119903 let 119873119886 = 0 Then we
obtain their inner layer information In 1198662119871 both 1198711198733119872119860119883 and1198711198734119872119860119883 are equal to 2 whereas 1198711198732119872119860119883 is equal to 1Overall the physical network is layered into six layers
with two inner layer networks associated with nodes N3 andN4The start node119873119903 is the only one in the first layer and theterminal node119873119886 is in three layers including L3 L5 and L6Thus we obtain three paths that can be used between119873119903 and119873119886 119871119875 is used to denote the length of path which is equal tothe number of VNFs that the path can hold LS is the lengthof an SFC which is equal to the VNF number of the SFCrequest
Algorithm 3 evaluates the nodes in the physical networkand selects a node which is most suitable to deploy therequired VNF After obtaining the layering informationAlgorithm 3makes the decision regarding whether themulti-ple links can satisfy the user request When themaximal layernumber in 119871119860 of the terminal nodes 119873119886 which is obtainedfrom the sumof all inner layers is smaller than the SFC length119871119878 in the user request the physical network cannot satisfy theuser request For example if we need to deploy an SFC intothe physical network in Figure 3(a) the start and terminalnodes are N1 and N9 The maximum layer number in 119871119860 is
Security and Communication Networks 7
Input (1) 119873119903 (2) 119873119886 (3) physical network GOutput 119866119871(1) 119873119903 997888rarr 1198811 119871119872119860119883 = L1(2) for 119881119871119872119860119883 = 0 119873119898 = 119873119886 do(3) for each119873119899 isin 119866 do(4) if 119873119898 larrrarr 119873119899 ampamp 119873119899 notin sum1198711198721198601198831 119881119883(5) 119873119899 997888rarr 119881119871119872119860119883+1 (6) else if 119873119898 larrrarr 119873119899 ampamp 119873119899 isin sum1198711198721198601198831 119881119883ampamp119873119899 =119873119886(7) 119873119899 997888rarr 119881119871119872119860119883+1 (8) end if(9) end for(10) 119871119872119860119883 ++(11) end for(12) for LX le 119871119872119860119883 do(13) for 119873119898 isin 119881119883 do(14) 119873119898 997888rarr 119881(1198831)(15) 119871119898119872119860119883 = 1198711198981(16) for 119881(119883119871119898119872119860119883) = 0 do(17) if 119873119899 isin 119881119883ampamp119873119898 larrrarr 119873119899ampamp119873119899 notin sum1198711198981198721198601198831 119866119883L(18) 119873119899 997888rarr 119881(119883119871119898119872119860119883)(19) end if(20) 119871119898119872119860119883 ++(21) end for(22) end for(23) end for(24) return 119866119871
Algorithm 2 Network layered processing
Input (1) SFC request(2) 119866119871(3) bool direction(4) X 119873119871 isin 119881119883Output the node119873119862 which has the minimum value of 120575(1) Temp = +infin(2) int i= 0(3) if direction is true(4) i= X-1(5) end if(6) if direction is false(7) i= X+1(8) end if(9) for 119873119898 isin 119881119894 do(10) if 119873119898 larrrarr 119873119871(11) if 119861119904119894 119898 gt 119861119903119894 119898ampamp119861119904119890 119898 gt 119861119903119890 119898 ampamp 119862119904 119898 gt 119862119903(12) Compute 120575 according to Equation (7)(13) if 120575 lt Temp(14) Temp = 120575(15) 119873119862 =119873119898(16) end if(17) end if(18) end if(19) end for(20) return 119873119862
Algorithm 3 Evaluation of related nodes
8 Security and Communication Networks
N1
N2
N3
N4 N6
N5
N8 N10
N9
N7
(a) Physical network
L1 V1 L3 V3L2 V2 L4 V4 L5 V5 L6 V6
N2
N3
N4
N5
N6
N9
N7
N8
N9
N10
N9N1
N4 N3N3 N4
6(21)3
6(22)3
6(21)4
6(22)4
(b) Processing procedure of layering
Figure 3 An example of a layered physical network
6 and there is an inner layer in layer L2The total sum is 7 sothis network topology can satisfy only an SFC request whoselength is no more than 7 An SFC request whose number ofVNFs ismore than 7 is too large for this network However aslong as the number of VNFs is not greater than the numberof nodes in the network our OSFCD-LSEM algorithm cantry to find more paths to deploy the longer SFC It may needadditional time and capacity of bandwidth resources becausethere are too many VNFs that need to be deployed Theproposed Algorithm 3 can search nodes in both positive andnegative directions When addressing an SFC that is longerthan the abovementioned sum Algorithm 3 commonly findsa suitable node in the layer119881119871119860+1 instead of in the upper layer119881119871119860minus1 then the algorithm can increase the length of the path(119871119875) However an extreme situation should be consideredsuch as when the node is in the last layer L119871119872119860119883 For thissituation our algorithm will find a node in the upper layer119881119871119872119860119883minus1 and run Algorithm 2 again
Finally we must evaluate the nodes from each layer ofthe layered network and select some nodes to map the VNFsAlgorithm 3 uses the abovementioned strategy to find a pathfrom 119873119886 to 119873119903 to host an SFC request and satisfy the userdemand Algorithm 3 selects the nodes from each layer using(7) and (8) The selected node must directly link to thenode in the next layer VN and must have sufficient resources
to deploy the VNFs and satisfy the corresponding functionrequirements
120575 = min(119861119904119894 119898 minus 119861119903119894 V119899119891119894119895119861119904119894 119898 119861119904119890 119898 minus 119861119903119890 V119899119891119894119895119861119904119890 119898 )
times 119862119904 119898 minus 119862119903 V119899119891119894119895119862119904 119898(7)
119862119904 119898 = 119862119905119900119905119886119897 119898 minus sum119878119894isin119878119865119862 119900119899119897119894119899119890
sumV119899119891119894119895isin119878119894
119862119903 V119899119891119894119895 times 120572V119899119891119894119895 119898 (8)
In (7) we use 120575 to denote the nodersquos appropriateness forthe VNF in the user request 119861119904119894 119898 is the idle bandwidth ofall links that connect node Nm with the nodes in the nextlayer 119881119873 and 119861119903119894 V119899119891119894119895 is the requested bandwidth resourcefrom vnf ij to the vnf ij+1 where V119899119891119894j is the j-th VNF in theSi (ie the i-th SFC request) 119861119904119890 119898 is the idle bandwidthof the path that connects node Nm with the nodes in theupper layer 119881119880 and 119861119903119890 V119899119891119894119895 is the requested bandwidthresource for making vnf ij connect with the last VNF 119862119904 119898represents the idle computing resources of node Nm and119862119903 V119899119891119894119895 represents the requested computing resources of thevnf ij In (8) Ctotal m represents the total computing resourcesof node Nm SFConline is the set of online SFC requests and
Security and Communication Networks 9
(a) US-Net topology (b) China-Net topology
Figure 4 Two real network topologies used in our simulations
120572V119899119891119894119895 119898 denotes whether vnf ij is deployed on node Nm Wecan evaluate whether a physical node is suitable to deploythe corresponding VNF according to the value of 120575 Afterphysical node evaluation we select the nodewith theminimalvalue of 120575 to hold the corresponding VNF
5 Simulation Results and Analysis
To evaluate the performance of our algorithm we compareour algorithm with two existing algorithms ie closed-loopwith critical mapping feedback (CCMF) [17] and key-VNFdeploy first (KVDF) [38] CCMF deploys the VNFs that havemore resource requirement priorities to optimize the totalconsumption of bandwidth resources KVDF focuses on therelation between different VNFs in the same SFC and therelation among different SFCs According to the relation andinfluence KVDF prioritizes the deployment of the key-VNFand key-SFC
To evaluate the performance of algorithms in differentnetwork scenarios we evaluate the performance of comparedalgorithms in moderate-scale and large-scale networks theUS-Network [50] (shown in Figure 4(a)) and the China-Network [51] (shown in Figure 4(b)) We use GT-ITM [52] togenerate the moderate-scale and large-scale network topolo-gies In moderate-scale network topologies there are approx-imately 100 nodes and 400 links In large-scale networktopologies there are approximately 300 nodes and 1000 linksTheUS-Net topology has 24 nodes and 43 links Additionallythe China-Net topology has 55 nodes and 103 links In thesenetwork topologies the bandwidth resource of all links isuniformly distributed at 100sim200 units and the computingresource of all nodes is set as 10 units
In our simulations we set the following parametersaccording to the existing work [49] The computer randomlygenerates user requests with lengths (ie Ls) from 5 to 14 andthese online SFC requests arrive as a Poisson process For eachphysical network and for a given 119871 119904 we randomly generate10000 SFC requests with the request client and destinationclient nodes which are randomly assigned to the physicalnodesThe computing resource demand of each VNF followsa uniform distribution U (1 3) and the bandwidth resource
demand of each virtual link follows a uniform distribution U(20 50)
With the increasing number of users and SFC requeststhe deployment of SFC in a static network becomes increas-ingly challenging Thus the network scalability must beimproved For a network-aware scaling strategy it is betterto extend the network instead of changing the network Innetwork G we define the evaluation of information 119866119878 in
119866119878 = 119871119872119860119883sum119883=1
sum119873119894isin119881119883
(119862119904 + 119861119904119894 + 119861119904119890) (9)
The OSFCD-LSEM algorithm layers the physical net-work obtains the layer which has minimum resourcesanalyzes its inner layer information and obtains the ldquoweakrdquonodes or links that influence the capacity of the networkThen the OSFCD-LSEM algorithm extends their resourcesto secure a more robust network
We obtained the simulation results by averaging theresults of multiple simulations We used an Ubuntu virtualmachine running on a computer with a 37 GHz Intel Corei3-4170 and 4 GB of RAM to run the simulations And thealgorithms were coded in Java programming language
From Figure 5 we can see the simulation results ofour OSFCD-LSEM algorithm in a moderate-scale networkFigure 5(a) shows the information of the physical networkWe can see that L8 limits the network capacity and will affectthe deployment of SFC Figure 5(b) shows the information ofthe physical nodes in L8 In L8 node-67 has the minimalamount of bandwidth and node-72 has the minimal amountof computing resources Both node-67 and node-72 are theldquoweak pointsrdquo in the physical network If we can increasetheir corresponding resources the capacity of the physicalnetwork will be enhanced For network operators increasingthe corresponding resources to the corresponding nodes andlinks is necessary and highly beneficial moreover they canalso obtain a more robust network
Figure 6 shows the simulation results on the US-Netnetwork Figure 6(a) shows the information about each layerin the overall network and we find that L4 may be theldquoweak pointrdquo of the physical network because of the lackof computing and bandwidth resources Figure 6(b) shows
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
6 Security and Communication Networks
Input (1) SFC request (2) physical network 119866Output SFC deployment scheme(1) Receive a user request(2) Initialize Path = [](3) 119873119886 997888rarr Path119873119871 =119873119886(4) Layer the topology Algorithm 2(119873119903119873119871 119866)(5) Get 119871119860 the layers that destination client belongs to(6) while 119871119878 gt max(119871119860) + sum119871119872119860119883119883=1 max(119871119894119872119860119883) forall119873119894 isin 119881119883 do(7) if max119871119860 = 119871119872119860119883(8) 119873119879119864119872119875 = Algorithm 3( true max119871119860)(9) 119873119879119864119872119875 997888rarr Path(10) 119873119871 =119873119879119864119872119875(11) else(12) 119873119879119864119872119875 = Algorithm 3( false max119871119860)(13) 119873119879119864119872119875 997888rarr Path(14) 119873119871=119873119879119864119872119875(15) end if(16) 119871119878 = 119871119878 - 1(17) VNF 997888rarr 119873119879119864119872119875(18) Algorithm 2(119873119903119873119871 119866)(19) Update 119871119860(20) end while(21) if 119871119878 lemax(119871119860)(22) Select min LX isin 119871119860 ampamp LX gt 119871119878(23) while 119873119903 notin Path do(24) 119873119879119864119872119875 = Algorithm 3 ( true LX)(25) 119873119879119864119872119875 997888rarr Path(26) 119873119871 =119873119879119864119872119875(27) LX = LX - 1(28) VNF997888rarr 119873119879119864119872119875(29) end while(30) end if(31) SFC deployment scheme is Path
Algorithm 1 OSFCD-LSEM algorithm
into L3 (we specify that all nodes can only belong to onelayer except the terminal node N9 thus node N4 cannot bepart of L3 even though it connects with node N3 which isin L2)
In our layered network except for the destination clientnode N9 the nodes in one layer must have links to the nodesin the previous layer Thus nodes N7 N8 and N10 directlyhave links to the nodes in L3 (ie nodes N5 N6 and N9)whereas N10 connects only with destination client node N9among the three nodes in L3 NodeN10 should not be placedin layer L4 Thus we place only nodes N7 and N8 into L4Nodes N9 and N10 connect with nodes N7 and N8 in L4thus we place nodes N9 and N10 in layer L5 and becauseN9 connects with node N10 our algorithm places node N9in L6 When ten nodes in G belong to the correspondinglayers the overall network-layered processing finishes Thusthe OSFCD-LSEM algorithm can guarantee finding a pathwithout loops For each LX we also need to layer and obtainthe inner layer 119866119883119871 In the example shown by Figure 3(b) thelayer L2 has an inner layer 1198662119871 that includes two layers For1198662119871 in Figure 3(b) each layer X le LMAX and each node Ni isinVX should be set as the start node 119873119903 let 119873119886 = 0 Then we
obtain their inner layer information In 1198662119871 both 1198711198733119872119860119883 and1198711198734119872119860119883 are equal to 2 whereas 1198711198732119872119860119883 is equal to 1Overall the physical network is layered into six layers
with two inner layer networks associated with nodes N3 andN4The start node119873119903 is the only one in the first layer and theterminal node119873119886 is in three layers including L3 L5 and L6Thus we obtain three paths that can be used between119873119903 and119873119886 119871119875 is used to denote the length of path which is equal tothe number of VNFs that the path can hold LS is the lengthof an SFC which is equal to the VNF number of the SFCrequest
Algorithm 3 evaluates the nodes in the physical networkand selects a node which is most suitable to deploy therequired VNF After obtaining the layering informationAlgorithm 3makes the decision regarding whether themulti-ple links can satisfy the user request When themaximal layernumber in 119871119860 of the terminal nodes 119873119886 which is obtainedfrom the sumof all inner layers is smaller than the SFC length119871119878 in the user request the physical network cannot satisfy theuser request For example if we need to deploy an SFC intothe physical network in Figure 3(a) the start and terminalnodes are N1 and N9 The maximum layer number in 119871119860 is
Security and Communication Networks 7
Input (1) 119873119903 (2) 119873119886 (3) physical network GOutput 119866119871(1) 119873119903 997888rarr 1198811 119871119872119860119883 = L1(2) for 119881119871119872119860119883 = 0 119873119898 = 119873119886 do(3) for each119873119899 isin 119866 do(4) if 119873119898 larrrarr 119873119899 ampamp 119873119899 notin sum1198711198721198601198831 119881119883(5) 119873119899 997888rarr 119881119871119872119860119883+1 (6) else if 119873119898 larrrarr 119873119899 ampamp 119873119899 isin sum1198711198721198601198831 119881119883ampamp119873119899 =119873119886(7) 119873119899 997888rarr 119881119871119872119860119883+1 (8) end if(9) end for(10) 119871119872119860119883 ++(11) end for(12) for LX le 119871119872119860119883 do(13) for 119873119898 isin 119881119883 do(14) 119873119898 997888rarr 119881(1198831)(15) 119871119898119872119860119883 = 1198711198981(16) for 119881(119883119871119898119872119860119883) = 0 do(17) if 119873119899 isin 119881119883ampamp119873119898 larrrarr 119873119899ampamp119873119899 notin sum1198711198981198721198601198831 119866119883L(18) 119873119899 997888rarr 119881(119883119871119898119872119860119883)(19) end if(20) 119871119898119872119860119883 ++(21) end for(22) end for(23) end for(24) return 119866119871
Algorithm 2 Network layered processing
Input (1) SFC request(2) 119866119871(3) bool direction(4) X 119873119871 isin 119881119883Output the node119873119862 which has the minimum value of 120575(1) Temp = +infin(2) int i= 0(3) if direction is true(4) i= X-1(5) end if(6) if direction is false(7) i= X+1(8) end if(9) for 119873119898 isin 119881119894 do(10) if 119873119898 larrrarr 119873119871(11) if 119861119904119894 119898 gt 119861119903119894 119898ampamp119861119904119890 119898 gt 119861119903119890 119898 ampamp 119862119904 119898 gt 119862119903(12) Compute 120575 according to Equation (7)(13) if 120575 lt Temp(14) Temp = 120575(15) 119873119862 =119873119898(16) end if(17) end if(18) end if(19) end for(20) return 119873119862
Algorithm 3 Evaluation of related nodes
8 Security and Communication Networks
N1
N2
N3
N4 N6
N5
N8 N10
N9
N7
(a) Physical network
L1 V1 L3 V3L2 V2 L4 V4 L5 V5 L6 V6
N2
N3
N4
N5
N6
N9
N7
N8
N9
N10
N9N1
N4 N3N3 N4
6(21)3
6(22)3
6(21)4
6(22)4
(b) Processing procedure of layering
Figure 3 An example of a layered physical network
6 and there is an inner layer in layer L2The total sum is 7 sothis network topology can satisfy only an SFC request whoselength is no more than 7 An SFC request whose number ofVNFs ismore than 7 is too large for this network However aslong as the number of VNFs is not greater than the numberof nodes in the network our OSFCD-LSEM algorithm cantry to find more paths to deploy the longer SFC It may needadditional time and capacity of bandwidth resources becausethere are too many VNFs that need to be deployed Theproposed Algorithm 3 can search nodes in both positive andnegative directions When addressing an SFC that is longerthan the abovementioned sum Algorithm 3 commonly findsa suitable node in the layer119881119871119860+1 instead of in the upper layer119881119871119860minus1 then the algorithm can increase the length of the path(119871119875) However an extreme situation should be consideredsuch as when the node is in the last layer L119871119872119860119883 For thissituation our algorithm will find a node in the upper layer119881119871119872119860119883minus1 and run Algorithm 2 again
Finally we must evaluate the nodes from each layer ofthe layered network and select some nodes to map the VNFsAlgorithm 3 uses the abovementioned strategy to find a pathfrom 119873119886 to 119873119903 to host an SFC request and satisfy the userdemand Algorithm 3 selects the nodes from each layer using(7) and (8) The selected node must directly link to thenode in the next layer VN and must have sufficient resources
to deploy the VNFs and satisfy the corresponding functionrequirements
120575 = min(119861119904119894 119898 minus 119861119903119894 V119899119891119894119895119861119904119894 119898 119861119904119890 119898 minus 119861119903119890 V119899119891119894119895119861119904119890 119898 )
times 119862119904 119898 minus 119862119903 V119899119891119894119895119862119904 119898(7)
119862119904 119898 = 119862119905119900119905119886119897 119898 minus sum119878119894isin119878119865119862 119900119899119897119894119899119890
sumV119899119891119894119895isin119878119894
119862119903 V119899119891119894119895 times 120572V119899119891119894119895 119898 (8)
In (7) we use 120575 to denote the nodersquos appropriateness forthe VNF in the user request 119861119904119894 119898 is the idle bandwidth ofall links that connect node Nm with the nodes in the nextlayer 119881119873 and 119861119903119894 V119899119891119894119895 is the requested bandwidth resourcefrom vnf ij to the vnf ij+1 where V119899119891119894j is the j-th VNF in theSi (ie the i-th SFC request) 119861119904119890 119898 is the idle bandwidthof the path that connects node Nm with the nodes in theupper layer 119881119880 and 119861119903119890 V119899119891119894119895 is the requested bandwidthresource for making vnf ij connect with the last VNF 119862119904 119898represents the idle computing resources of node Nm and119862119903 V119899119891119894119895 represents the requested computing resources of thevnf ij In (8) Ctotal m represents the total computing resourcesof node Nm SFConline is the set of online SFC requests and
Security and Communication Networks 9
(a) US-Net topology (b) China-Net topology
Figure 4 Two real network topologies used in our simulations
120572V119899119891119894119895 119898 denotes whether vnf ij is deployed on node Nm Wecan evaluate whether a physical node is suitable to deploythe corresponding VNF according to the value of 120575 Afterphysical node evaluation we select the nodewith theminimalvalue of 120575 to hold the corresponding VNF
5 Simulation Results and Analysis
To evaluate the performance of our algorithm we compareour algorithm with two existing algorithms ie closed-loopwith critical mapping feedback (CCMF) [17] and key-VNFdeploy first (KVDF) [38] CCMF deploys the VNFs that havemore resource requirement priorities to optimize the totalconsumption of bandwidth resources KVDF focuses on therelation between different VNFs in the same SFC and therelation among different SFCs According to the relation andinfluence KVDF prioritizes the deployment of the key-VNFand key-SFC
To evaluate the performance of algorithms in differentnetwork scenarios we evaluate the performance of comparedalgorithms in moderate-scale and large-scale networks theUS-Network [50] (shown in Figure 4(a)) and the China-Network [51] (shown in Figure 4(b)) We use GT-ITM [52] togenerate the moderate-scale and large-scale network topolo-gies In moderate-scale network topologies there are approx-imately 100 nodes and 400 links In large-scale networktopologies there are approximately 300 nodes and 1000 linksTheUS-Net topology has 24 nodes and 43 links Additionallythe China-Net topology has 55 nodes and 103 links In thesenetwork topologies the bandwidth resource of all links isuniformly distributed at 100sim200 units and the computingresource of all nodes is set as 10 units
In our simulations we set the following parametersaccording to the existing work [49] The computer randomlygenerates user requests with lengths (ie Ls) from 5 to 14 andthese online SFC requests arrive as a Poisson process For eachphysical network and for a given 119871 119904 we randomly generate10000 SFC requests with the request client and destinationclient nodes which are randomly assigned to the physicalnodesThe computing resource demand of each VNF followsa uniform distribution U (1 3) and the bandwidth resource
demand of each virtual link follows a uniform distribution U(20 50)
With the increasing number of users and SFC requeststhe deployment of SFC in a static network becomes increas-ingly challenging Thus the network scalability must beimproved For a network-aware scaling strategy it is betterto extend the network instead of changing the network Innetwork G we define the evaluation of information 119866119878 in
119866119878 = 119871119872119860119883sum119883=1
sum119873119894isin119881119883
(119862119904 + 119861119904119894 + 119861119904119890) (9)
The OSFCD-LSEM algorithm layers the physical net-work obtains the layer which has minimum resourcesanalyzes its inner layer information and obtains the ldquoweakrdquonodes or links that influence the capacity of the networkThen the OSFCD-LSEM algorithm extends their resourcesto secure a more robust network
We obtained the simulation results by averaging theresults of multiple simulations We used an Ubuntu virtualmachine running on a computer with a 37 GHz Intel Corei3-4170 and 4 GB of RAM to run the simulations And thealgorithms were coded in Java programming language
From Figure 5 we can see the simulation results ofour OSFCD-LSEM algorithm in a moderate-scale networkFigure 5(a) shows the information of the physical networkWe can see that L8 limits the network capacity and will affectthe deployment of SFC Figure 5(b) shows the information ofthe physical nodes in L8 In L8 node-67 has the minimalamount of bandwidth and node-72 has the minimal amountof computing resources Both node-67 and node-72 are theldquoweak pointsrdquo in the physical network If we can increasetheir corresponding resources the capacity of the physicalnetwork will be enhanced For network operators increasingthe corresponding resources to the corresponding nodes andlinks is necessary and highly beneficial moreover they canalso obtain a more robust network
Figure 6 shows the simulation results on the US-Netnetwork Figure 6(a) shows the information about each layerin the overall network and we find that L4 may be theldquoweak pointrdquo of the physical network because of the lackof computing and bandwidth resources Figure 6(b) shows
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 7
Input (1) 119873119903 (2) 119873119886 (3) physical network GOutput 119866119871(1) 119873119903 997888rarr 1198811 119871119872119860119883 = L1(2) for 119881119871119872119860119883 = 0 119873119898 = 119873119886 do(3) for each119873119899 isin 119866 do(4) if 119873119898 larrrarr 119873119899 ampamp 119873119899 notin sum1198711198721198601198831 119881119883(5) 119873119899 997888rarr 119881119871119872119860119883+1 (6) else if 119873119898 larrrarr 119873119899 ampamp 119873119899 isin sum1198711198721198601198831 119881119883ampamp119873119899 =119873119886(7) 119873119899 997888rarr 119881119871119872119860119883+1 (8) end if(9) end for(10) 119871119872119860119883 ++(11) end for(12) for LX le 119871119872119860119883 do(13) for 119873119898 isin 119881119883 do(14) 119873119898 997888rarr 119881(1198831)(15) 119871119898119872119860119883 = 1198711198981(16) for 119881(119883119871119898119872119860119883) = 0 do(17) if 119873119899 isin 119881119883ampamp119873119898 larrrarr 119873119899ampamp119873119899 notin sum1198711198981198721198601198831 119866119883L(18) 119873119899 997888rarr 119881(119883119871119898119872119860119883)(19) end if(20) 119871119898119872119860119883 ++(21) end for(22) end for(23) end for(24) return 119866119871
Algorithm 2 Network layered processing
Input (1) SFC request(2) 119866119871(3) bool direction(4) X 119873119871 isin 119881119883Output the node119873119862 which has the minimum value of 120575(1) Temp = +infin(2) int i= 0(3) if direction is true(4) i= X-1(5) end if(6) if direction is false(7) i= X+1(8) end if(9) for 119873119898 isin 119881119894 do(10) if 119873119898 larrrarr 119873119871(11) if 119861119904119894 119898 gt 119861119903119894 119898ampamp119861119904119890 119898 gt 119861119903119890 119898 ampamp 119862119904 119898 gt 119862119903(12) Compute 120575 according to Equation (7)(13) if 120575 lt Temp(14) Temp = 120575(15) 119873119862 =119873119898(16) end if(17) end if(18) end if(19) end for(20) return 119873119862
Algorithm 3 Evaluation of related nodes
8 Security and Communication Networks
N1
N2
N3
N4 N6
N5
N8 N10
N9
N7
(a) Physical network
L1 V1 L3 V3L2 V2 L4 V4 L5 V5 L6 V6
N2
N3
N4
N5
N6
N9
N7
N8
N9
N10
N9N1
N4 N3N3 N4
6(21)3
6(22)3
6(21)4
6(22)4
(b) Processing procedure of layering
Figure 3 An example of a layered physical network
6 and there is an inner layer in layer L2The total sum is 7 sothis network topology can satisfy only an SFC request whoselength is no more than 7 An SFC request whose number ofVNFs ismore than 7 is too large for this network However aslong as the number of VNFs is not greater than the numberof nodes in the network our OSFCD-LSEM algorithm cantry to find more paths to deploy the longer SFC It may needadditional time and capacity of bandwidth resources becausethere are too many VNFs that need to be deployed Theproposed Algorithm 3 can search nodes in both positive andnegative directions When addressing an SFC that is longerthan the abovementioned sum Algorithm 3 commonly findsa suitable node in the layer119881119871119860+1 instead of in the upper layer119881119871119860minus1 then the algorithm can increase the length of the path(119871119875) However an extreme situation should be consideredsuch as when the node is in the last layer L119871119872119860119883 For thissituation our algorithm will find a node in the upper layer119881119871119872119860119883minus1 and run Algorithm 2 again
Finally we must evaluate the nodes from each layer ofthe layered network and select some nodes to map the VNFsAlgorithm 3 uses the abovementioned strategy to find a pathfrom 119873119886 to 119873119903 to host an SFC request and satisfy the userdemand Algorithm 3 selects the nodes from each layer using(7) and (8) The selected node must directly link to thenode in the next layer VN and must have sufficient resources
to deploy the VNFs and satisfy the corresponding functionrequirements
120575 = min(119861119904119894 119898 minus 119861119903119894 V119899119891119894119895119861119904119894 119898 119861119904119890 119898 minus 119861119903119890 V119899119891119894119895119861119904119890 119898 )
times 119862119904 119898 minus 119862119903 V119899119891119894119895119862119904 119898(7)
119862119904 119898 = 119862119905119900119905119886119897 119898 minus sum119878119894isin119878119865119862 119900119899119897119894119899119890
sumV119899119891119894119895isin119878119894
119862119903 V119899119891119894119895 times 120572V119899119891119894119895 119898 (8)
In (7) we use 120575 to denote the nodersquos appropriateness forthe VNF in the user request 119861119904119894 119898 is the idle bandwidth ofall links that connect node Nm with the nodes in the nextlayer 119881119873 and 119861119903119894 V119899119891119894119895 is the requested bandwidth resourcefrom vnf ij to the vnf ij+1 where V119899119891119894j is the j-th VNF in theSi (ie the i-th SFC request) 119861119904119890 119898 is the idle bandwidthof the path that connects node Nm with the nodes in theupper layer 119881119880 and 119861119903119890 V119899119891119894119895 is the requested bandwidthresource for making vnf ij connect with the last VNF 119862119904 119898represents the idle computing resources of node Nm and119862119903 V119899119891119894119895 represents the requested computing resources of thevnf ij In (8) Ctotal m represents the total computing resourcesof node Nm SFConline is the set of online SFC requests and
Security and Communication Networks 9
(a) US-Net topology (b) China-Net topology
Figure 4 Two real network topologies used in our simulations
120572V119899119891119894119895 119898 denotes whether vnf ij is deployed on node Nm Wecan evaluate whether a physical node is suitable to deploythe corresponding VNF according to the value of 120575 Afterphysical node evaluation we select the nodewith theminimalvalue of 120575 to hold the corresponding VNF
5 Simulation Results and Analysis
To evaluate the performance of our algorithm we compareour algorithm with two existing algorithms ie closed-loopwith critical mapping feedback (CCMF) [17] and key-VNFdeploy first (KVDF) [38] CCMF deploys the VNFs that havemore resource requirement priorities to optimize the totalconsumption of bandwidth resources KVDF focuses on therelation between different VNFs in the same SFC and therelation among different SFCs According to the relation andinfluence KVDF prioritizes the deployment of the key-VNFand key-SFC
To evaluate the performance of algorithms in differentnetwork scenarios we evaluate the performance of comparedalgorithms in moderate-scale and large-scale networks theUS-Network [50] (shown in Figure 4(a)) and the China-Network [51] (shown in Figure 4(b)) We use GT-ITM [52] togenerate the moderate-scale and large-scale network topolo-gies In moderate-scale network topologies there are approx-imately 100 nodes and 400 links In large-scale networktopologies there are approximately 300 nodes and 1000 linksTheUS-Net topology has 24 nodes and 43 links Additionallythe China-Net topology has 55 nodes and 103 links In thesenetwork topologies the bandwidth resource of all links isuniformly distributed at 100sim200 units and the computingresource of all nodes is set as 10 units
In our simulations we set the following parametersaccording to the existing work [49] The computer randomlygenerates user requests with lengths (ie Ls) from 5 to 14 andthese online SFC requests arrive as a Poisson process For eachphysical network and for a given 119871 119904 we randomly generate10000 SFC requests with the request client and destinationclient nodes which are randomly assigned to the physicalnodesThe computing resource demand of each VNF followsa uniform distribution U (1 3) and the bandwidth resource
demand of each virtual link follows a uniform distribution U(20 50)
With the increasing number of users and SFC requeststhe deployment of SFC in a static network becomes increas-ingly challenging Thus the network scalability must beimproved For a network-aware scaling strategy it is betterto extend the network instead of changing the network Innetwork G we define the evaluation of information 119866119878 in
119866119878 = 119871119872119860119883sum119883=1
sum119873119894isin119881119883
(119862119904 + 119861119904119894 + 119861119904119890) (9)
The OSFCD-LSEM algorithm layers the physical net-work obtains the layer which has minimum resourcesanalyzes its inner layer information and obtains the ldquoweakrdquonodes or links that influence the capacity of the networkThen the OSFCD-LSEM algorithm extends their resourcesto secure a more robust network
We obtained the simulation results by averaging theresults of multiple simulations We used an Ubuntu virtualmachine running on a computer with a 37 GHz Intel Corei3-4170 and 4 GB of RAM to run the simulations And thealgorithms were coded in Java programming language
From Figure 5 we can see the simulation results ofour OSFCD-LSEM algorithm in a moderate-scale networkFigure 5(a) shows the information of the physical networkWe can see that L8 limits the network capacity and will affectthe deployment of SFC Figure 5(b) shows the information ofthe physical nodes in L8 In L8 node-67 has the minimalamount of bandwidth and node-72 has the minimal amountof computing resources Both node-67 and node-72 are theldquoweak pointsrdquo in the physical network If we can increasetheir corresponding resources the capacity of the physicalnetwork will be enhanced For network operators increasingthe corresponding resources to the corresponding nodes andlinks is necessary and highly beneficial moreover they canalso obtain a more robust network
Figure 6 shows the simulation results on the US-Netnetwork Figure 6(a) shows the information about each layerin the overall network and we find that L4 may be theldquoweak pointrdquo of the physical network because of the lackof computing and bandwidth resources Figure 6(b) shows
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
8 Security and Communication Networks
N1
N2
N3
N4 N6
N5
N8 N10
N9
N7
(a) Physical network
L1 V1 L3 V3L2 V2 L4 V4 L5 V5 L6 V6
N2
N3
N4
N5
N6
N9
N7
N8
N9
N10
N9N1
N4 N3N3 N4
6(21)3
6(22)3
6(21)4
6(22)4
(b) Processing procedure of layering
Figure 3 An example of a layered physical network
6 and there is an inner layer in layer L2The total sum is 7 sothis network topology can satisfy only an SFC request whoselength is no more than 7 An SFC request whose number ofVNFs ismore than 7 is too large for this network However aslong as the number of VNFs is not greater than the numberof nodes in the network our OSFCD-LSEM algorithm cantry to find more paths to deploy the longer SFC It may needadditional time and capacity of bandwidth resources becausethere are too many VNFs that need to be deployed Theproposed Algorithm 3 can search nodes in both positive andnegative directions When addressing an SFC that is longerthan the abovementioned sum Algorithm 3 commonly findsa suitable node in the layer119881119871119860+1 instead of in the upper layer119881119871119860minus1 then the algorithm can increase the length of the path(119871119875) However an extreme situation should be consideredsuch as when the node is in the last layer L119871119872119860119883 For thissituation our algorithm will find a node in the upper layer119881119871119872119860119883minus1 and run Algorithm 2 again
Finally we must evaluate the nodes from each layer ofthe layered network and select some nodes to map the VNFsAlgorithm 3 uses the abovementioned strategy to find a pathfrom 119873119886 to 119873119903 to host an SFC request and satisfy the userdemand Algorithm 3 selects the nodes from each layer using(7) and (8) The selected node must directly link to thenode in the next layer VN and must have sufficient resources
to deploy the VNFs and satisfy the corresponding functionrequirements
120575 = min(119861119904119894 119898 minus 119861119903119894 V119899119891119894119895119861119904119894 119898 119861119904119890 119898 minus 119861119903119890 V119899119891119894119895119861119904119890 119898 )
times 119862119904 119898 minus 119862119903 V119899119891119894119895119862119904 119898(7)
119862119904 119898 = 119862119905119900119905119886119897 119898 minus sum119878119894isin119878119865119862 119900119899119897119894119899119890
sumV119899119891119894119895isin119878119894
119862119903 V119899119891119894119895 times 120572V119899119891119894119895 119898 (8)
In (7) we use 120575 to denote the nodersquos appropriateness forthe VNF in the user request 119861119904119894 119898 is the idle bandwidth ofall links that connect node Nm with the nodes in the nextlayer 119881119873 and 119861119903119894 V119899119891119894119895 is the requested bandwidth resourcefrom vnf ij to the vnf ij+1 where V119899119891119894j is the j-th VNF in theSi (ie the i-th SFC request) 119861119904119890 119898 is the idle bandwidthof the path that connects node Nm with the nodes in theupper layer 119881119880 and 119861119903119890 V119899119891119894119895 is the requested bandwidthresource for making vnf ij connect with the last VNF 119862119904 119898represents the idle computing resources of node Nm and119862119903 V119899119891119894119895 represents the requested computing resources of thevnf ij In (8) Ctotal m represents the total computing resourcesof node Nm SFConline is the set of online SFC requests and
Security and Communication Networks 9
(a) US-Net topology (b) China-Net topology
Figure 4 Two real network topologies used in our simulations
120572V119899119891119894119895 119898 denotes whether vnf ij is deployed on node Nm Wecan evaluate whether a physical node is suitable to deploythe corresponding VNF according to the value of 120575 Afterphysical node evaluation we select the nodewith theminimalvalue of 120575 to hold the corresponding VNF
5 Simulation Results and Analysis
To evaluate the performance of our algorithm we compareour algorithm with two existing algorithms ie closed-loopwith critical mapping feedback (CCMF) [17] and key-VNFdeploy first (KVDF) [38] CCMF deploys the VNFs that havemore resource requirement priorities to optimize the totalconsumption of bandwidth resources KVDF focuses on therelation between different VNFs in the same SFC and therelation among different SFCs According to the relation andinfluence KVDF prioritizes the deployment of the key-VNFand key-SFC
To evaluate the performance of algorithms in differentnetwork scenarios we evaluate the performance of comparedalgorithms in moderate-scale and large-scale networks theUS-Network [50] (shown in Figure 4(a)) and the China-Network [51] (shown in Figure 4(b)) We use GT-ITM [52] togenerate the moderate-scale and large-scale network topolo-gies In moderate-scale network topologies there are approx-imately 100 nodes and 400 links In large-scale networktopologies there are approximately 300 nodes and 1000 linksTheUS-Net topology has 24 nodes and 43 links Additionallythe China-Net topology has 55 nodes and 103 links In thesenetwork topologies the bandwidth resource of all links isuniformly distributed at 100sim200 units and the computingresource of all nodes is set as 10 units
In our simulations we set the following parametersaccording to the existing work [49] The computer randomlygenerates user requests with lengths (ie Ls) from 5 to 14 andthese online SFC requests arrive as a Poisson process For eachphysical network and for a given 119871 119904 we randomly generate10000 SFC requests with the request client and destinationclient nodes which are randomly assigned to the physicalnodesThe computing resource demand of each VNF followsa uniform distribution U (1 3) and the bandwidth resource
demand of each virtual link follows a uniform distribution U(20 50)
With the increasing number of users and SFC requeststhe deployment of SFC in a static network becomes increas-ingly challenging Thus the network scalability must beimproved For a network-aware scaling strategy it is betterto extend the network instead of changing the network Innetwork G we define the evaluation of information 119866119878 in
119866119878 = 119871119872119860119883sum119883=1
sum119873119894isin119881119883
(119862119904 + 119861119904119894 + 119861119904119890) (9)
The OSFCD-LSEM algorithm layers the physical net-work obtains the layer which has minimum resourcesanalyzes its inner layer information and obtains the ldquoweakrdquonodes or links that influence the capacity of the networkThen the OSFCD-LSEM algorithm extends their resourcesto secure a more robust network
We obtained the simulation results by averaging theresults of multiple simulations We used an Ubuntu virtualmachine running on a computer with a 37 GHz Intel Corei3-4170 and 4 GB of RAM to run the simulations And thealgorithms were coded in Java programming language
From Figure 5 we can see the simulation results ofour OSFCD-LSEM algorithm in a moderate-scale networkFigure 5(a) shows the information of the physical networkWe can see that L8 limits the network capacity and will affectthe deployment of SFC Figure 5(b) shows the information ofthe physical nodes in L8 In L8 node-67 has the minimalamount of bandwidth and node-72 has the minimal amountof computing resources Both node-67 and node-72 are theldquoweak pointsrdquo in the physical network If we can increasetheir corresponding resources the capacity of the physicalnetwork will be enhanced For network operators increasingthe corresponding resources to the corresponding nodes andlinks is necessary and highly beneficial moreover they canalso obtain a more robust network
Figure 6 shows the simulation results on the US-Netnetwork Figure 6(a) shows the information about each layerin the overall network and we find that L4 may be theldquoweak pointrdquo of the physical network because of the lackof computing and bandwidth resources Figure 6(b) shows
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 9
(a) US-Net topology (b) China-Net topology
Figure 4 Two real network topologies used in our simulations
120572V119899119891119894119895 119898 denotes whether vnf ij is deployed on node Nm Wecan evaluate whether a physical node is suitable to deploythe corresponding VNF according to the value of 120575 Afterphysical node evaluation we select the nodewith theminimalvalue of 120575 to hold the corresponding VNF
5 Simulation Results and Analysis
To evaluate the performance of our algorithm we compareour algorithm with two existing algorithms ie closed-loopwith critical mapping feedback (CCMF) [17] and key-VNFdeploy first (KVDF) [38] CCMF deploys the VNFs that havemore resource requirement priorities to optimize the totalconsumption of bandwidth resources KVDF focuses on therelation between different VNFs in the same SFC and therelation among different SFCs According to the relation andinfluence KVDF prioritizes the deployment of the key-VNFand key-SFC
To evaluate the performance of algorithms in differentnetwork scenarios we evaluate the performance of comparedalgorithms in moderate-scale and large-scale networks theUS-Network [50] (shown in Figure 4(a)) and the China-Network [51] (shown in Figure 4(b)) We use GT-ITM [52] togenerate the moderate-scale and large-scale network topolo-gies In moderate-scale network topologies there are approx-imately 100 nodes and 400 links In large-scale networktopologies there are approximately 300 nodes and 1000 linksTheUS-Net topology has 24 nodes and 43 links Additionallythe China-Net topology has 55 nodes and 103 links In thesenetwork topologies the bandwidth resource of all links isuniformly distributed at 100sim200 units and the computingresource of all nodes is set as 10 units
In our simulations we set the following parametersaccording to the existing work [49] The computer randomlygenerates user requests with lengths (ie Ls) from 5 to 14 andthese online SFC requests arrive as a Poisson process For eachphysical network and for a given 119871 119904 we randomly generate10000 SFC requests with the request client and destinationclient nodes which are randomly assigned to the physicalnodesThe computing resource demand of each VNF followsa uniform distribution U (1 3) and the bandwidth resource
demand of each virtual link follows a uniform distribution U(20 50)
With the increasing number of users and SFC requeststhe deployment of SFC in a static network becomes increas-ingly challenging Thus the network scalability must beimproved For a network-aware scaling strategy it is betterto extend the network instead of changing the network Innetwork G we define the evaluation of information 119866119878 in
119866119878 = 119871119872119860119883sum119883=1
sum119873119894isin119881119883
(119862119904 + 119861119904119894 + 119861119904119890) (9)
The OSFCD-LSEM algorithm layers the physical net-work obtains the layer which has minimum resourcesanalyzes its inner layer information and obtains the ldquoweakrdquonodes or links that influence the capacity of the networkThen the OSFCD-LSEM algorithm extends their resourcesto secure a more robust network
We obtained the simulation results by averaging theresults of multiple simulations We used an Ubuntu virtualmachine running on a computer with a 37 GHz Intel Corei3-4170 and 4 GB of RAM to run the simulations And thealgorithms were coded in Java programming language
From Figure 5 we can see the simulation results ofour OSFCD-LSEM algorithm in a moderate-scale networkFigure 5(a) shows the information of the physical networkWe can see that L8 limits the network capacity and will affectthe deployment of SFC Figure 5(b) shows the information ofthe physical nodes in L8 In L8 node-67 has the minimalamount of bandwidth and node-72 has the minimal amountof computing resources Both node-67 and node-72 are theldquoweak pointsrdquo in the physical network If we can increasetheir corresponding resources the capacity of the physicalnetwork will be enhanced For network operators increasingthe corresponding resources to the corresponding nodes andlinks is necessary and highly beneficial moreover they canalso obtain a more robust network
Figure 6 shows the simulation results on the US-Netnetwork Figure 6(a) shows the information about each layerin the overall network and we find that L4 may be theldquoweak pointrdquo of the physical network because of the lackof computing and bandwidth resources Figure 6(b) shows
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
10 Security and Communication Networks
1 2 3 4 5 6 7 8 9 10 11 12 130100200300400500600700800900
10001100
Capa
city
(uni
ts)
Layer Node resource capacityLink resource capacity
(a) The resource capacity of each layer
61 65 67 69 720
20406080
100120140160180200
Capa
city
(uni
ts)
Node Computing resource Bandwidth resource of last layerBandwidth resource of next layer
(b) The information of the physical nodes in L8
Figure 5 Simulation results for the moderate-scale network
1 2 3 4 5 60
50
100
150
200
250
300
350
Capa
city
(uni
ts)
Layer
Node resource capacity Link resource capacity
(a) The resource capacity of each layer
5 9 10 130
5
10
15
20
25
30
35
40Ca
paci
ty (u
nits)
Node
Computing resourceBandwidth resource of last layer Bandwidth resource of next layer
(b) The detailed information of nodes in L4
Figure 6 Simulation results for the US-Net network
the nodesrsquo detailed information about L4 In the evaluationresult node 5 has no computing resources to carry any VNFThismeans that node 5 is a ldquodead pointrdquo in the network and itsignificantly compromises network performance Moreoverthere are a few bandwidth resources in node 5 for connectingthe nodes in the last layer and the next layer It is necessaryand urgent to supplement the corresponding bandwidthresources to make the network more robust With respectto node 13 though there are enough computing resourcesthere are insufficient bandwidth resources Thus increasingbandwidth resources in node 5 to connect the nodes in L3and L4 will significantly improve the ability of the physicalnetwork to provision more SFC requests and make thenetwork more robust
For network operators and our OSFCD-LSEM algorithmthose ldquoweak pointsrdquo serve as the focal points for the extensionof the physical network Relative to an approach that blindly
extends the physical network to all nodes and links withoutfocus and direction considerations the extension imple-mented by theOSFCD-LSEM algorithm ismore accurate andefficient Our OSFCD-LSEM algorithm can be used to supplyresources to the nodes that require the most resources and itavoids wasting resources while improving the total physicalnetwork capacity
Figure 7 shows the simulation results of the acceptanceratios of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 7(a) 7(b) 7(c) and 7(d) show thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively The OSFCD-LSEM algorithm has a higher SFC request acceptance ratiothan the CCMF and KVDF algorithms in all networks As aresult the OSFCD-LSEM can find the appropriate nodes forVNF deployment based on the clientrsquos direction with the helpof the network layering algorithm Compared with the other
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 11
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120Ac
cept
ance
Rat
io (
)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
140
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
20
40
60
80
100
120
Acce
ptan
ce R
atio
()
Length of SFC
SFCD-LEMB CCMF KVDF(d) Results for China-Net network
Figure 7 Acceptance ratios in different physical networks
two algorithms OSFCD-LSEM always finds the shortest andmost appropriate path to deploy SFCs thus it has the highestacceptance ratio among the algorithms Furthermore theOSFCD-LSEM algorithm has a relatively stable acceptanceratio in arbitrary scale networks and different 119871 119904 becauseour OSFCD-LSEM algorithm evaluates the network afterlayering part of the network and can appropriately finish theSFC deployments In addition the OSFCD-LSEM algorithmperforms better in both networks that are generated by GT-ITM and in the real networks The excellent performance isachieved because it is based on the evaluation of the layerednetwork
Figure 8 shows the simulation results of the runningtime of the OSFCD-LSEM algorithm and the other twoalgorithms Figures 8(a) 8(b) 8(c) and 8(d) reveal thecomparable results in the moderate-scale large-scale US-Net and China-Net networks respectively Among the threealgorithms the running time of the OSFCD-LSEM algorithmfor performing the deployment is shortest because it can findnodes for VNF deployment based on the appropriate searchdirection rather than on random searching Therefore the
OSFCD-LSEM can find the path towards the client nodewithin fewer searching steps and shorter searching timecomparedwith the other algorithmsMoreover the optimiza-tion of the OSFCD-LSEM algorithm makes its running timeslowly increase with the increasing length of SFC (119871 119904) As 119871 119904increases OSFCD-LSEM requires only a few searching stepsand the running time slowly increases due to the directionalsearch advantage
Figures 9(a) 9(b) 9(c) and 9(d) show the simulationresults of the consumptions of bandwidth resources in themoderate-scale large-scale US-Net and China-Net net-works respectively Figure 9 shows that the OSFCD-LSEMalgorithm can provision SFC requests with less bandwidthconsumption than the other two algorithms for the sameSFC requests Since it employs the network layering strategythe shortest paths can be found by the OSFCD-LSEMalgorithm to deploy the SFC based on an efficient searchdirection Then compared with the other two algorithmsSFC communication via the shortest paths consumes lessbandwidth resources With the increase in 119871 119904 and net-work size the OSFCD-LSEM algorithm shows outstanding
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
12 Security and Communication Networks
5 6 7 8 9 10 11 12 13 1402468
1012141618202224262830
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
102030405060708090
100110
Runn
ing
Tim
e (un
its)
Length of SFC
SFCD-LEMB CCMF KCDF
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
1
2
3
4
5
6
Runn
ing
Tim
e (un
its)
Length of SFC SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 8 Running time in different physical networks
performance in saving bandwidth resources because itobtains the layering information of the nodes and links inthe network by layered strategies which is one of the maincontributions of OSFCD-LSEM Importantly since savingbandwidth resources is the goal of this study the simula-tions of different network scenarios and SFC lengths (119871 119904)show that the OSFCD-LSEM algorithm obtains the expectedresults
In summary the OSFCD-LSEM algorithm layers thephysical network and is superior to the other two comparedalgorithms in terms of the acceptance ratio running timeand bandwidth consumption These achievements can beattributed to the manner in which the OSFCD-LSEM algo-rithm obtains overall evaluation information of the physicalnetwork
6 Conclusion
Under NFV technology the network functions can bemigrated from dedicated hardware and be deployed ontocommercial servers in any necessary location NFV can
remarkably enhance the flexibility and reduce the resourceconsumption in the physical network With the benefit ofNFV the clientsrsquo experience and network performance areboth improved
In this paper we study the efficient online social IoTapplication oriented SFC provision problem We propose anSFC deployment algorithm OSFCD-LSEM which layers thephysical network to obtain the layering information of thenodes and links in the network Moreover it also selects themost suitable node to host the VNFs by evaluating somenodes in the physical network The simulation results showthat the OSFCD-LSEM algorithm has better performancein terms of time efficiency acceptance ratio and bandwidthconsumption for provisioning social IoT application orientedSFC requests Furthermore to satisfy the increasing socialIoT demands we can use the layering information to appro-priately extend the physical network
In the future work we can introduce artificial intelligencealgorithm to the existing framework to improve the accuracyrate or to study the algorithm to run efficiently in a morecomplex physical network environment
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 13
5 6 7 8 9 10 11 12 13 140
50
100
150
200
250
300
350
400
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(a) Results for moderate-scale network
5 6 7 8 9 10 11 12 13 140
40
80
120
160
200
240
280
320
Length of SFC SFCD-LEMB CCMF KVDF
Band
wid
th C
onsu
mpt
ion
(uni
ts)
(b) Results for large-scale network
5 6 7 8 9 10 11 12 13 140
100200300400500600700800900
1000
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(c) Results for US-Net network
5 6 7 8 9 10 11 12 13 140
306090
120150180210240270300330
Band
wid
th C
onsu
mpt
ion
(uni
ts)
Length of SFC
SFCD-LEMB CCMF KVDF
(d) Results for China-Net network
Figure 9 Bandwidth consumptions in different physical networks
Data Availability
The data supporting the results reported in this work isgenerated in our simulation experiments in our study
Conflicts of Interest
The authors declare that they have no conflicts of interest
Acknowledgments
This research was partially supported by the National NaturalScience Foundation of China (61571098) Sichuan Scienceand Technology Program (2019YFG0206) and the 111 Project(B14039)
References
[1] M S Yoon and A E Kamal ldquoNFV resource allocation usingmixed queuing network modelrdquo in Proceedings of the IEEEGlobal Communications Conference pp 1ndash7 2016
[2] G Sun S Sun J Sun H Yu X Du and M GuizanildquoSecurity and privacy preservation in fog-based crowd sensing
on the internet of vehiclesrdquo Journal of Network and ComputerApplications vol 134 pp 89ndash99 2019
[3] G Sun V Chang M Ramachandran et al ldquoEfficient locationprivacy algorithm for Internet of Things (IoT) services andapplicationsrdquo Journal of Network and Computer Applicationsvol 89 pp 3ndash13 2017
[4] Q Du H Song and X Zhu ldquoSocial-feature enabled communi-cations among devices toward the smart iot communityrdquo IEEECommunications Magazine vol 57 no 1 pp 130ndash137 2019
[5] Y Dai D Xu S Maharjan and Y Zhang ldquoJoint computationoffloading and user association in multi-task mobile edgecomputingrdquo IEEE Transactions on Vehicular Technology vol 67no 12 pp 12313ndash12325 2018
[6] G Sun D Liao H Li H Yu and V Chang ldquoL2P2 A location-label based approach for privacy preserving in LBSrdquo FutureGeneration Computer Systems vol 74 pp 375ndash384 2017
[7] G Sun L Song D Liao H Yu andV Chang ldquoTowards privacypreservation for ldquocheck-inrdquo services in location-based socialnetworksrdquo Information Sciences vol 481 pp 616ndash634 2019
[8] H Song G A Fink and S Jeschke Security and Privacy inCyber-Physical Systems Foundations Principles and Applica-tions Wiley-IEEE Press 2017
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
14 Security and Communication Networks
[9] X Huang Y Lu D Li and MMa ldquoA novel mechanism for fastdetection of transformed data leakagerdquo IEEE Access vol 6 pp35926ndash35936 2018
[10] G Sun Y Xie D Liao H Yu and V Chang ldquoUser-defined pri-vacy location-sharing system inmobile online social networksrdquoJournal of Network and Computer Applications vol 86 pp 34ndash45 2017
[11] R Cohen L Lewin-Eytan J S Naor andD Raz ldquoNear optimalplacement of virtual network functionsrdquo in Proceedings of theIEEE Conference on Computer Communications pp 1346ndash13542015
[12] B Addis D Belabed M Bouet and S Secci ldquoVirtual networkfunctions placement and routing optimizationrdquo in Proceedingsof the IEEE 4th International Conference on Cloud Networkingpp 171ndash177 2015
[13] J Liu W Lu F Zhou P Lu and Z Zhu ldquoOn Dynamicservice function chain deployment and readjustmentrdquo IEEETransactions on Network and Service Management vol 14 no3 pp 543ndash553 2017
[14] S MehraghdamM Keller andH Karl ldquoSpecifying and placingchains of virtual network functionsrdquo in Proceedings of the IEEE3rd International Conference on Cloud Networking pp 7ndash132014
[15] X Wang C Wu F Le et al ldquoOnline VNF scaling in datacen-tersrdquo pp 1-9 2016
[16] G Sun G Zhu D Liao H Yu X Du and M Guizani ldquoCost-efficient service function chain orchestration for low-latencyapplications in NFV networksrdquo IEEE Systems Journal pp 1ndash13
[17] Z Ye X Cao J Wang H Yu and C Qiao ldquoJoint topologydesign and mapping of service function chains for efficientscalable and reliable network functions virtualizationrdquo IEEENetwork vol 30 no 3 pp 81ndash87 2016
[18] S Clayman E Maini A Galis et al ldquoThe dynamic placementof virtual network functionsrdquo IEEE Network Operations andManagement Symposiu pp 1ndash9 2014
[19] G Sun V Chang G Yang and D Liao ldquoThe cost-efficientdeployment of replica servers in virtual content distributionnetworks for data fusionrdquo Information Sciences vol 432 pp495ndash515 2018
[20] P S Khodashenas B Blanco M-A Kourtis et al ldquoServicemapping and orchestration over multi-tenant cloud-enabledRANrdquo IEEE Transactions on Network and Service Managementvol 14 no 4 pp 904ndash919 2017
[21] Y Li and M Chen ldquoSoftware-defined network function virtu-alization A surveyrdquo IEEE Access vol 3 pp 2542ndash2553 2015
[22] X Du and H H Chen ldquoSecurity in wireless sensor networksrdquoIEEEWireless Communications Magazine vol 15 no 4 pp 60ndash66 2008
[23] X Du Y Xiao M Guizani and H-H Chen ldquoAn effective keymanagement scheme for heterogeneous sensor networksrdquo AdHoc Networks vol 5 no 1 pp 24ndash34 2007
[24] H Song R Srinivasan T Sookoor et al Smart Cities Founda-tions Principles and Applications Wiley 2017
[25] X Du M Guizani Y Xiao et al ldquoA routing-driven ellipticcurve cryptography based key management scheme for het-erogeneous sensor networksrdquo IEEE Transactions on WirelessCommunications vol 8 no 3 pp 1223ndash1229 2009
[26] L Liu C Chen T Qiu M Zhang S Li and B Zhou ldquoA datadissemination scheme based on clustering and probabilisticbroadcasting in VANETsrdquo Vehicular Communications vol 13pp 78ndash88 2018
[27] X Li and C Qian ldquoLow-complexity multi-resource packetscheduling for network function virtualizationrdquo in Proceedingsof the IEEEConference onComputer Communications pp 1400ndash1408 2015
[28] G Sun Y Li D Liao and V Chang ldquoService function chainorchestration across multiple domains A full mesh aggregationapproachrdquo IEEE Transactions on Network and Service Manage-ment vol 15 no 3 pp 1175ndash1191 2018
[29] G Xiong Y-X Hu L Tian J-L Lan J-F Li and Q Zhou ldquoAvirtual service placement approach based on improved quan-tum genetic algorithmrdquo Frontiers of Information Technology andElectronic Engineering vol 17 no 7 pp 661ndash671 2016
[30] G Sun Y Li Y Li D Liao and V Chang ldquoLow-latencyorchestration for workflow-oriented service function chain inedge computingrdquo Future Generation Computer Systems vol 85pp 116ndash128 2018
[31] G Sun D Liao D Zhao Z Xu and H Yu ldquoLive migrationfor multiple correlated virtual machines in cloud-based datacentersrdquo IEEE Transactions on Services Computing vol 11 no2 pp 279ndash291 2018
[32] V Eramo E Miucci M Ammar and F G Lavacca ldquoAnapproach for service function chain routing and virtual func-tion network instance migration in network function virtual-ization architecturesrdquo IEEEACM Transactions on Networkingvol 25 no 4 pp 2008ndash2025 2017
[33] R Mijumbi J Serrat J-L Gorricho et al ldquoNetwork functionvirtualization state-of-the-art and research challengesrdquo IEEECommunications Surveys amp Tutorials vol 18 no 1 pp 236ndash2622015
[34] W Cerroni and F Callegati ldquoLive migration of virtual networkfunctions in cloud-based edge networksrdquo in Proceedings of theIEEE International Conference on Communications pp 2963ndash2968 2014
[35] Y Sang B Ji G R Gupta X Du and L Ye ldquoProvably efficientalgorithms for joint placement and allocation of virtual networkfunctionsrdquo in Proceedings of the IEEE Conference on ComputerCommunications pp 829ndash837 2017
[36] J Batalle J F Riera E Escalona and J A Garcıa-Espın ldquoOnthe implementation of NFV over an OpenFlow infrastructureRouting function virtualizationrdquo Software Defined Networks forFuture Networks and Services pp 1ndash6 2013
[37] W Ma C Medina and D Pan ldquoTraffic-aware placement ofNFV middleboxesrdquo in Proceedings of the IEEE Global Commu-nications Conference pp 1ndash6 2015
[38] Q Sun P LuW Lu and Z Zhu ldquoForecast-assistedNFV servicechain deployment based on affiliation-aware vNF placementrdquo inProceedings of the IEEE Global Communications Conference pp1ndash6 2017
[39] L Qu C Assi and K Shaban ldquoDelay-aware scheduling andresource optimization with network function virtualizationrdquoIEEE Transactions on Communications vol 64 no 9 pp 3746ndash3758 2016
[40] S Herker X An W Kiess S Beker and A Kirstaedter ldquoData-center architecture impacts on virtualized network functionsservice chain embedding with high availability requirementsrdquoin Proceedings of the IEEE Global Communications Conferencepp 1ndash7 2015
[41] T-W Kuo B-H Liou K C-J Lin and M-J Tsai ldquoDeployingchains of virtual network functions On the relation betweenlink and server usagerdquo in Proceedings of the IEEE Conference onComputer Communications pp 1ndash9 2016
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
Security and Communication Networks 15
[42] H Jin D Pan J Xu et al ldquoEfficient VM placement withmultiple deterministic and stochastic resources in data centersrdquoin Proceedings of the IEEE Global Communications Conferencepp 2505ndash2510 2012
[43] M T Beck and J F Botero ldquoCoordinated allocation of servicefunction chainsrdquo in Proceedings of the IEEEGlobal Communica-tions Conference pp 1ndash6 2015
[44] Q Zhang X Wang I Kim P Palacharla and T IkeuchildquoVertex-centric computation of service function chains inmulti-domain networksrdquo in Proceedings of the IEEE Conferenceon Network Softwarization pp 211ndash218 2016
[45] T Wen H Yu G Sun et al ldquoNetwork Function Consolidationin Service Function Chaining Orchestrationrdquo in Proceedings ofthe IEEE International Conference on Communications pp 1ndash62016
[46] T Park Y Kim J Park H Suh B Hong and S Shin ldquoQoSEQuality of security a network security framework with dis-tributed NFVrdquo in Proceedings of the IEEE International Confer-ence on Communications pp 1ndash6 2016
[47] Y Li L T X Phan andB T Loo ldquoNetwork functions virtualiza-tion with soft real-time guaranteesrdquo in Proceedings of the IEEEConference on Computer Communications pp 1ndash9 2016
[48] F Z Yousaf P Loureiro F Zdarsky T Taleb and M LiebschldquoCost analysis of initial deployment strategies for virtualizedmobile core network functionsrdquo IEEE Communications Maga-zine vol 53 no 12 pp 60ndash66 2015
[49] G Sun Y Li H Yu A V Vasilakos X Du and M GuizanildquoEnergy-efficient and traffic-aware service function chainingorchestration in multi-domain networksrdquo Future GenerationComputer Systems vol 91 pp 347ndash360 2019
[50] A Haque and P-H Ho ldquoA study on the design of survivableoptical virtual private networks (O-VPN)rdquo IEEE Transactionson Reliability vol 55 no 3 pp 516ndash524 2006
[51] X Xie N Hua and X Zheng ldquoDynamic routing wavelengthand core allocation in multi-core fiber based optical networksconsidering inter-core crosstalkrdquo in Proceedings of the SignalProcessing in Photonic Communications (ppJM3A-4) OpticalSociety of America 2015
[52] K Calvert J Eagan S Merugu A Namjoshi J Stasko and EZegura ldquoExtending and enhancing GT-ITMrdquo in Proceedings ofthe ACM SigcommWorkshop on Models pp 23ndash27 2003
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom
International Journal of
AerospaceEngineeringHindawiwwwhindawicom Volume 2018
RoboticsJournal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Active and Passive Electronic Components
VLSI Design
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Shock and Vibration
Hindawiwwwhindawicom Volume 2018
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawiwwwhindawicom
Volume 2018
Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom
The Scientific World Journal
Volume 2018
Control Scienceand Engineering
Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom
Journal ofEngineeringVolume 2018
SensorsJournal of
Hindawiwwwhindawicom Volume 2018
International Journal of
RotatingMachinery
Hindawiwwwhindawicom Volume 2018
Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawiwwwhindawicom Volume 2018
Hindawiwwwhindawicom Volume 2018
Navigation and Observation
International Journal of
Hindawi
wwwhindawicom Volume 2018
Advances in
Multimedia
Submit your manuscripts atwwwhindawicom