Distributed Association Method Assisted by Cell for...

Research ArticleDistributed Association Method Assisted by Cell forEfficiency Enhancement of Wireless Networks

Jaesung Park

Department of Information Security University of Suwon San 2-2 Wau-ri Bongdam-eup Hwaseong-siGyeonggi-do 445-743 Republic of Korea

Correspondence should be addressed to Jaesung Park jaesungparksuwonackr

Received 22 November 2016 Accepted 24 January 2017 Published 16 February 2017

Academic Editor Hideyuki Takahashi

Copyright copy 2017 Jaesung ParkThis 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

In this paper we propose a distributed cell association scheme called cell-guided association method (CGAM) to improve theefficiency of awireless network InCGAMMSs attempt to associatewith their best cellsHowever unlike the conventionalmethodscells do not passively accept the association requests of MSs Instead a cell determines whether to accept an association requestor not by considering the performance of MSs already associated with it and that of the requesting MS If a cell cannot provide acertain level of service to them it rejects the association request and guides the requesting MS to select another cell that gives thenext maximum performance metric to the MS Since our method takes the cell resource usage into consideration it can increasethe resource efficiency of a wireless network while enhancing the overall data rate provided to MSs by balancing the number ofMSs in a cell Through performance comparisons by simulation studies we verify that CGAM outperforms maximum SINR-basedmethod and QoS-based method in terms of the total data rate provided by a system and outage probabilities of MSs

1 Introduction

As the number of handheld devices proliferates fast mobiledata traffic has increased tremendously over a few decadesThe traffic demand is expected to grow over the futurebecause of the advent of Internet of things [1 2] To copewith the traffic demands wireless networks becomemore andmore complex and the importance of distributed resourcemanagement methods that increase both the resource effi-ciency of a network and quality of service (QoS) experiencesby users attracts much attention Among those since a cellassociation policy that matches a cell and a mobile station(MS) affects the distribution of MSs it influences systemefficiency and QoS of users

Traditionally MSs attempt to associate with a cell thatgive them the strongest signal-to-interference and noise ratio(SINR) The rationale behind this is that data rate providedto a MS enhances as the signal quality between a MS and acell increases [3] However the data rate provided to a MS isdetermined not only by the SINR but also by the amount ofradio resources allocated to it AMS canmeasure SINRs fromits adjacent cells However aMSdoes not know the amount of

resources that it can obtain froma cell before it associateswiththe cell Since cell resources are shared by the MSs associatedwith a cell the fraction of resources allocated to each MSdecreases as the number of MSs in a cell increasesThereforeif a MS selects a cell to associate with based on the SINRit may not receive the highest data rate In addition SINRbetween a cell and a MS is determined by the transmit powerof a cell the level of interference and the random fadingBecause of the randomness it becomes highly probable thatthe number of MSs in a cell is not evenly distributed amongcells even if cells are carefully deployed considering the trafficdemands of areas covered by them If the load unbalanceamong cells occurs MSs in a lightly populated cell enjoy highdata rate while MSs in a densely populated cell suffer fromlow data rate because of the resource contention amongMSs

To overcome this problem cell load-aware or QoS-oriented cell associationmethods have been proposed In [4]joint optimization of user association policies and antennatilts settings is proposed by predicting cell load accuratelyThe authors in [5] propose two distributed association meth-ods that enable MSs to make association decisions based onthe information gathered by probing neighboring cells In [6]

HindawiMobile Information SystemsVolume 2017 Article ID 9701267 7 pageshttpsdoiorg10115520179701267

2 Mobile Information Systems

a use association policy is proposed to minimize the energyconsumption in LTE access networks which are composedof small cells deployed densely However cell load-awareassociationmethods require that cells broadcast the cell loadsperiodically or on-demand In addition MSs select a cell toassociate with to maximize its benefit selfishly while cellspassively accept the association request fromMSsThereforesystem resources are not used in an optimal way because ofthe selfishness

To address this issue we devise a distributed cell-guidedassociation method (CGAM) In CGAM cells do not pas-sively accept the association requests from MSs On the con-trary cells determine whether or not to accept the associationrequest based on their resource availabilityThus even if MSsoperate selfishly to maximize its payoff system can guideMSs to achieve its own objective such as increasing resourceutilization and cell load balance In addition since CGAMdoes not require cells to broadcasts cell load conventional celloperations need not be changed

The rest of the paper is organized as follows In Section 2we review related works on association management InSection 3 we describe the CGAM in detail after introducingsystemmodel In Section 4we present anddiscuss simulationresults to verify CGAM by comparing the performances ofCGAM and SINR-based association method and cell load-aware association method

2 Related Works

Since the cell association problem is to find a set of MSsand cells pairs that optimize a given objective function ofa wireless network optimization methods are widely usedto design an association rule Various objective functionsare devised for the cell association problem In [7] authorspropose an algorithm that maximizes the system revenuewhile associating MSs with the minimum total transmissionpower They use Bendersrsquo decomposition to solve nonconvexoptimization problem optimally Throughput maximizationis also used as an objective function In [8] sum rate ismaximized and authors in [9 10] maximize the log-utility ofa network under the proportional fairness In [11] MSs areassociated with BSs in a way that global outage probability isminimizedHowever it has been shown that the optimizationproblem is an NP-hard problem

To overcome the NP-hardness different kinds of opti-mization techniques are used to solve the cell associationproblem Markov decision process is used in [12 13] How-ever it is hard to define state transition probability andthe complexity increases substantially as the number ofstates increases In [14] the structure of the network utilitymaximization problem is used to solve the problem directlyDual decomposition is often employed to design a distributedalgorithm by relaxing the optimization constraints [10 15]However since the computational cost of these methods ishigh it is not clear whether these methods can be applied ata small time scale when the values of input variables changevery fast

Game theory is also applied to solve the cell associationproblem In [16] an evolutionary game theory is used

to devise an algorithm for cell association and antennaallocation in 5G networks with massive MIMO Authorsin [17] show that the game that users selfishly select basestations giving them the best throughput and BSs allocatethe same time to their users has one Nash equilibrium pointwhich achieves proportional fairness system-wide Howeverthe convergence of the algorithms based on a game theoryis not guaranteed generally Additionally in terms of theimplementation they usually cause large overhead whichdeteriorates resource utilization

Stochastic geometry has been used to analyze the cellassociation problem In [18] authors propose a distributedbelief propagation algorithm to resolve user association prob-lem and analyze the average sparsity and degree distributionusing stochastic geometry Stochastic geometry is also usedin [19] to analyze the service success probability Then theyderive the impact of cell association and user schedulingon the service success probability Unlike the optimizationmethods that maximize the utility function for the currentnetwork configuration stochastic geometry performs opti-mization over the average utility Therefore even though thecomplexity and overhead of a stochastic geometry approachare lower than those of repeating optimization processwhenever network configuration changes the results will besuboptimal

Unlike the association methods based on a theoreticalstatic interference model a practical measurement basedinterference-aware association policy is proposed in [20]

There have been research works that optimize jointlycell association and other performance metrics In [21] atractable framework is proposed to analyze the performanceof eICIC by jointly considering cell association resourcepartitioning and transmit power reduction An energy effi-cient user association scheme is proposed in [22 23] JointBS association and power control algorithm that updatesiteratively BS association solution then the transmit power ofeach user is proposed in [24]

3 Cell-Guided Association Method

31 System Model We consider a downlink of a wireless net-work that provides best effort data service to MSs We denoteby119873119888 the set of cells in a network and by |119873119888| the cardinalityof119873119888 We assume that the system fully reuses frequency

Since radio resources of a cell are allocated to MSs inthe unit of radio resource block (RB) in modern wirelessnetworks such as LTE we assume that each cell 119909 has 119877119909 RBsWe consider long term SINR and throughput In other wordswe assume that the timescale measuring these values is muchlarger than that of fast channel variation Thus if we denotethe transmit power of a cell 119909 as 119875119909 the SINR between a MS119910 and a cell 119909 is given by

119878119909119910 =119875119909119866119909119910

119860119873 + sum119888isin119873119888minus119909 119875119888119866119888119910 (1)

where 119860119873 is the noise power and 119866119909119910 denotes the channelgain between a cell 119909 and a MS 119910 which includes the system

Mobile Information Systems 3

parameter such as antenna gain and channel parameters suchas path loss and shadowing

The spectral efficiency is an increasing function of 119878119909119910A variety of functions have been devised to reflect theeffect of operational frequency and modulation and codingschemes [25ndash27] However to avoid system dependency weuse Shannonrsquos formula to obtain the data rate for given SINRIf a cell 119909 allocates 119877119909119910 RBs to a MS 119910 the data rate of 119910 isobtained as

119863119909119910 = 119877119909119910119882119877log2 (1 + 119878119909119910) (2)

where 119882119877 is the bandwidth of a RB The number of RBsallocated to aMS is determined by the type of scheduler a celluses A scheduler considers the number of MSs sharing cellresources the channel quality between a cell and a MS whenit allocates resources at each frame In case of a round-robinscheduler the long term RBs allocated to each MS becomes119877119909|119872119909| where119872119909 is the set of MSs that a cell 119909 serves

However it is shown that there is a multiuser diversitygain when a cell exploits a proportional fair scheduler [28ndash30] In this paper we assume a cell uses a proportionalfair scheduler Then according to the results of [30] 119877119909119910 isdetermined as follows

119877119909119910 =1198771199091003816100381610038161003816119872119909

1003816100381610038161003816119892 (119872119909) (3)

where 119892(119872119909) = sum|119872119909|119894=1 (1119894) is the multiuser diversity gain

32QoSMetric When aMS requests for a guaranteed servicesuch as a voice call the number of RBs to provide the serviceis guaranteed by a cell Moreover no more RBs are allocatedto the request even if a cell has surplus RBs However whena cell provides data service the number of RBs allocated toeach MS varies according to the number of MSs served bya cell simultaneously For example even when there is onlyone MS associated with a cell the cell allocates all its RBsto the MS We also note that even if a MS uses data serviceusers tend to give up using a network when they did notreceive a minimum rate from a network Thus we use theoutage probability [31ndash33] as the QoS metric for a MS Theoutage probability is defined as the probability that a datarate provided to a MS is less than a given threshold 120579119889 Weuse the outage probability model derived in a network whosefrequency reuse factor is one and where rayleigh fadingchannel is assumed [33] When a MS 119910 associates with a cell119909 the outage probability of 119910 is modeled as

119874119909119910 = Prob (119863119909119910 lt 120579119889)

= 1

minus 119890

119860119873119861119909119910119866119909119910119875119909 prod

119894isin119873119888minus119909

1 (119866119894119910119875119894)

119861119909119910 (119866119909119910119875119909) + 1 (119866119894119910119875119894)

(4)

where 119861119909119910 = 2120579119889119863119909119910 minus 1 Since 119874119909119910 is a decreasing functionof119863119909119910119874119909119910 decreases if119877119909119910 or 119878119909119910 increasesTherefore119874119909119910can be regarded as a metric that represents the degree of cellload

33 CGAM Algorithm CGAM is composed of two algo-rithms a MS algorithm and a cell algorithm that operateindependently of each other Algorithm 1 shows each algo-rithm

A MS that needs to associate with a cell searches neigh-boring cells by hearing signals sent by them Then a MS 119910constructs a set of adjacent cells (119862119910) that contains the cellidentification numbers and SINR received from each cell in119862119910 Then 119910 initializes a potential cell list 1198621015840119910 = 119862119910 andattempts to associate with a cell 119909 in 1198621015840119910 that gives it themaximum SINR by sending an association request messageand waits for the response from 119909 If 119910 receives an associationresponse message from 119909 saying that 119909 does not accept therequest 119910 removes the cell 119909 in 1198621015840119910 and attempts to associatewith a cell in 1198621015840119910 giving the highest SINR to it (ie the secondbest cell in 119862119910 in terms of SINR) A MS 119910 repeats the sameprocess until it receives a positive association response froma cell If 119910 does not receive a positive response from any cellin 119862119910 it means that 119910 cannot be provided with the minimumdata rate from its neighboring cells Thus it suspend theassociation attempt

Upon receiving an association requestmessage from aMS119910 a cell 119909 calculates the outage probabilities of 119910 and 119895 in119872119909 by assuming that it accepts the request If none of theoutage probabilities are below a threshold value 120579119900 119909 acceptsthe association request of 119910 by sending a positive associationresponse message to 119910 Otherwise 119909 guides 119910 to associatewith other cells that may satisfy the outage probabilityconstraint by sending a negative association response to 119910

4 Performance Evaluation

In this section we evaluate the performance of the proposedcell association methods Specifically we compare the perfor-mance of the following three methods in terms of the outageprobability of a MS and data rates provided by cells

(i) SNM a MS associates with a cell according to themaximum signal strength and a cell does not guidea MS to associate with better cell

(ii) ONM a MS associates with a cell based on theminimumoutage probability and a cell does not guidea MS to associate with better cell

(iii) SOM a MS associates with a cell based on themaximum signal strength and a cell guide a MS toassociate with better cell with the outage probabilityit can provide to an MS

We construct urban macrocell topology following thescenario specified in 3GPP [27]We deployed 18 base stationsEach base station has three sectors We assume a hexagonalcell and each cell has 6 neighboring cells The systembandwidth is set to 5MHz and the bandwidth of a RB isconfigured as 180KHz The intersite distance is set to 500mTransmit power of a base station is configured to be 46 dBmAntenna gain of a base station is 45 dBi and that of a MSis 2 dBi The noise power is configured as minus11145 dBm andfrequency reuse factor is set to 1


MS Algorithm(0) cell search(1) construct Cy(2) 1198621015840119910 larr 119862119910(3)While 1198621015840119910 = (4) 119909 larr argmax 119894 isin 1198621015840119910 (119878119894119910)(5) send association request message to 119909(6) receive association responsemessage from 119909(7) if (response = OK)(8) 1198621015840119910 larr 1198621015840119910 minus 119909Cell Algorithm(0) Upon receiving association request from a MS 119910(1)1198721015840119909 larr119872119909 cup 119910(2) flaglarr 0(3) while 1198721015840119909 = (4) calculate 119874119909119895 (119895 isin 1198721015840119909)(5) if (119874119909119895 gt 120579119900)(6) send association response with NOK(7) flaglarr 1(8) break(9) else(10) 1198721015840119909 larr119872119909 minus 119894(11) if (flag == 0)(12) send association response with OK

Algorithm 1 Cell-guided association method

The path lossmodel of 1281+376 log10(max(119889 0035)) isapplied where 119889 is the distance in km between a sender and areceiver We assume log-normal shadowing channel that haszero mean and standard deviation of 120590dB = 8 dB

We locate a MS in a network randomly following theuniform distribution and apply each method for a MS todetermine a cell to associate with All the statistics aregathered when the number of MSs reach 5000

41 MS Performance To quantify the performance ofMS weuse two metrics One metric is the data rate received by a MSand the other is the outage probability of a MS According tothe locations of MSs we further classify MSs into edge MSswhich are defined as the MSs whose Euclidean distance fromits serving base station is larger than intersite distance overthree

Figure 1 shows the cumulative distribution of data ratesreceived by MSs and Figure 2 shows the cumulative distribu-tion of outage probabilities experienced by MSs ComparingSNMwith ONM performances of MSs increase by changingcell selection metric from SINR to outage probability Forexample 70 MSs receive less than 5758 Kbps when SINRis used as a cell selection metric On the contrary 70 MSsreceive 6079 Kbps when MSs use outage probability to selecta cell In terms of the outage probability the proportion ofMSswhose outage probabilities are less than 02 is 0794whenSINR is used On the contrary the proportion increases to0817 when MS selects a cell with an outage probability

The performances of MSs increase dramatically when acell guidesMSswhen they select a cell to associatewithWhen

Table 1 Cell statistics in terms of the number of MSs in a cell

SNM ONM SOMFI 088 096 092Avg 215 208 137Std 80 41 43

the association method is changed from SNM to SOM thedata rate received by 70 MSs increases from 5758 Kbps to9042 Kbps In addition the proportion of MSs whose outageprobabilities are less than 02 increases 106 times from 0794to 0842

The performance gain obtained by guiding MSs withoutage probability is more significant for the edge MSsNinety percentage of edge MSs receive less than 7933 Kbpswhen SNM is used while it increases to 13019 Kbps whenSOM is used The outage probabilities of 90 edge MSsare less than 05 when SNM is applied while the numberdecreases to 037 when MSs use SOM to associate with a cell

42 System Performance Figure 3 shows the total data rateprovided by each cell called cell rate We can observe thatthe cell rate increase dramatically if there is a guide by a cellThis is attributed to the fact that if MSs choose cells selfishlyto maximize their own performance metric some cells arecrowded with MSs while the others are loosely populated Toexamine the distribution of MSs in a cell in Table 1 we showJainrsquos fairness index (FI) in terms of the number of MSs in acell In the table we also show the average and the standard


0

01

02

03

04

05

06

07

08

09

1

0 500 1000 1500 2000

Prop

ortio

n of

MSs

Data rates of MSs (Kbps)SNMONMSOM

(a) All MSs

0

01

02

03

04

05

06

07

08

09

1

0 500 1000 1500 2000

Prop

ortio

n of

MSs

Data rates of edge MSs (Kbps)SNMONMSOM

(b) Edge MSs

Figure 1 Cumulative distribution of data rates received by MSs

0

01

02

03

04

05

06

07

08

09

1

0 005 01 015 02 025 03 035 04 045 05

Prop

ortio

n of

MSs

Outage prob of MSsSNMONMSOM

(a) All MSs

0

01

02

03

04

05

06

07

08

09

1

0 01 02 03 04 05 06 07

Prop

ortio

n of

MSs

Outage prob of edge MSs

SNMONMSOM

(b) Edge MSs

Figure 2 Cumulative distribution of MS outage probability

deviation of the number of MSs in a cell We can find thatFI is 088 and the standard deviation is 80 when SNM isused while FI and the standard deviation of SOM are 092and 43 respectively which means that MSs are more evenlypopulated when SOM is used than when SNM is used ONMalso distribute MSs evenly However the average number ofMSs in a cell of ONM is higher than those of SOM Since cellresources are shared among MSs the number of RBs that isallocated to eachMS in a cell becomes smaller as the numberof MSs in a cell increases In addition since SINR is inverselyproportional to the distance from a base station and aMS and

we deployed MSs according to the uniform distribution theaverage data rate per RB decreases as the number of MSs ina cell increases Therefore even though cell rate of ONM ishigher than that of SNM it is smaller than the cell rate ofSOM

To further investigate the benefit of cell guidance weshow the minimum data rate provided to a MS by each cellin Figure 4 and the maximum outage probability among theMSs in each cell experience in Figure 5 We can observe thatthe minimum data rate provided to MSs in a cell increasesand the maximum outage probability in each cell decreases


708090

100110120130140

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM

Figure 3 Total data rate provided by a cell

0005

01015

02025

03035

04045

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM

Figure 4 Minimum data rate provided to a MS associated with acell

0010203040506070809

1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Out

age p

rob

Cell IDSNMONMSOM

Figure 5Maximum outage probability provided to aMS associatedwith a cell

5 Conclusions

In this paper we propose a cell-guided association methodUnlike other cell load aware methods where MSs estimatethe load of neighboring cells before attempting to makeassociations with a cell in CGAM a cell decides whether ornot to accept an association request from a MS Consideringthe amount of available resources a cell rejects the associationrequest if it cannot provide the minimum data rate to a MSThe rejection guides a MS to try to associate with the second

best cellThus even if aMS selfishly selects a cell to maximizeits benefit they are not densely populated in some cells whilethe other cells are sparsely populated Therefore CGAMincreases not only the resource utilization of a system but alsothe quality of service perceived by MSs In addition CGAMdoes not require additional information in a cell broadcastmessage and conventional cell selection metric of a MS

Competing Interests

The author declares that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was partly supported by the Basic ScienceResearch Program through the National Research Founda-tion of Korea (NRF) funded by the Ministry of Education(NRF-2015R1D1A1A01060117) and by the GRRC programof Gyeonggi Province [(GRRC SUWON2016-B2) Study onObject Recognition System for Industrial Security and Intel-ligent Control System for Establishing Social SafetyNet Basedon Advanced Neuro-Fuzzy Technologies]

References

[1] D Evans ldquoThe Internet of Things How the Next Evolution ofthe Internet is Changing Everythingrdquo white paper April 2011

[2] Cisco ldquoCisco Visual Networking Index Global Mobile DataTraffic Forecast Update 2014ndash2019rdquoWhite Paper February 2015

[3] J G Andrews S Singh Q Ye X Lin and H S Dhillon ldquoAnoverview of load balancing in HetNets old myths and openproblemsrdquo IEEEWireless Communications vol 21 no 2 pp 18ndash25 2014

[4] A J Fehske H Klessig J Voigt and G P Fettweis ldquoConcurrentload-aware adjustment of user association and antenna tilts inself-organizing radio networksrdquo IEEE Transactions onVehicularTechnology vol 62 no 5 pp 1974ndash1988 2013

[5] K C Garikipati and K G Shin ldquoDistributed associationcontrol in shared wireless networksrdquo in Proceedings of the 10thAnnual IEEE Communications Society Conference on Sensingand Communication inWireless Networks (SECON rsquo13) pp 362ndash370 New Orleans La USA June 2013

[6] C Bottai C Cicconetti A Morelli M Rosellini and C VitaleldquoEnergy-efficient user association in extremely dense smallcell networksrdquo in Proceedings of the European Conference onNetworks and Communications (EuCNC rsquo14) pp 1ndash5 BolognaItaly June 2014

[7] L P Qian Y J A Zhang YWu and J Chen ldquoJoint base stationassociation and power control via Bendersrsquo decompositionrdquoIEEE Transactions on Wireless Communications vol 12 no 4pp 1651ndash1665 2013

[8] S Corroy L Falconetti and R Mathar ldquoCell association insmall heterogeneous networks downlink sum rate andmin ratemaximizationrdquo in Proceedings of the IEEE Wireless Communi-cations and Networking Conference (WCNC rsquo12) pp 888ndash892Paris France April 2012

[9] K Shen andW Yu ldquoDistributed pricing-based user associationfor downlink heterogeneous cellular networksrdquo IEEE Journal on


Selected Areas in Communications vol 32 no 6 pp 1100ndash11132014

[10] Q Ye B Rong Y Chen M Al-Shalash C Caramanis and J GAndrews ldquoUser association for load balancing in heterogeneouscellular networksrdquo IEEE Transactions on Wireless Communica-tions vol 12 no 6 pp 2706ndash2716 2013

[11] H Boostanimehr and V K Bhargava ldquoUnified and distributedQoS-driven cell association algorithms in heterogeneous net-worksrdquo IEEE Transactions on Wireless Communications vol 14no 3 pp 1650ndash1662 2015

[12] D Kumar E Altman and J-M Keli ldquoGlobally optimal user-Network Association in an 80211 WLAN amp 3G UMTS hybridcellrdquo in Proceedings of the 20th International Teletraffic Confer-ence on Managing Traffic Performance in Converged Networkspp 1173ndash1187 Ottawa Canada June 2007

[13] C Sun E Stevens-Navarro and V W S Wong ldquoA constrainedMDP-based vertical handoff decision algorithm for 4G wirelessnetworksrdquo in Proceedings of the IEEE International Conferenceon Communications (ICC rsquo08) pp 2169ndash2174 IEEE BeijingChina May 2008

[14] K Son S Chong and G De Veciana ldquoDynamic associationfor load balancing and interference avoidance in multi-cellnetworksrdquo IEEE Transactions on Wireless Communications vol8 no 7 pp 3566ndash3576 2009

[15] K Chitti Q Kuang and J Speidel ldquoJoint base station associa-tion and power allocation for uplink sum-rate maximizationrdquoin Proceedings of the IEEE 14th Workshop on Signal ProcessingAdvances in Wireless Communications (SPAWC rsquo13) pp 6ndash10Darmstadt Germany June 2013

[16] PWangW Song D Niyato and Y Xiao ldquoQoS-aware cell asso-ciation in 5G heterogeneous networks with massive MIMOrdquoIEEE Network vol 29 no 6 pp 76ndash82 2015

[17] L Jiang S Parekh and J Walrand ldquoBase station associationgame inmulti-cell wireless networksrdquo in Proceedings of the IEEEWireless Communications and Networking Conference (WCNCrsquo08) pp 1616ndash1621 Las Vegas Nev USA March 2008

[18] Y Chen J Li Z Lin G Mao and B Vucetic ldquoUser asso-ciation with unequal user priorities in heterogeneous cellularnetworksrdquo IEEE Transactions on Vehicular Technology vol 65no 9 pp 7374ndash7388 2016

[19] S Chen M Peng H Zhang and C Wang ldquoInvestigation ofservice success probability for downlink heterogeneous cellularnetworks with cell association and user schedulingrdquo in Pro-ceedings of the IEEE Wireless Communications and NetworkingConference (WCNC rsquo15) pp 1434ndash1439 IEEE New Orleans LaUSA March 2015

[20] B Rengarajan and G De Veciana ldquoPractical adaptive user asso-ciation policies for wireless systems with dynamic interferencerdquoIEEEACM Transactions on Networking vol 19 no 6 pp 1690ndash1703 2011

[21] W Tang S Feng Y Liu and M C Reed ldquoJoint low-powertransmit and cell association in heterogeneous networksrdquo inProceedings of the 58th IEEEGlobal Communications Conference(GLOBECOM rsquo15) pp 1ndash6 IEEE San Diego Calif USADecember 2015

[22] Y Zhu Z Zeng T Zhang L An and L Xiao ldquoAn energy effi-cient user association scheme based on cell sleeping in LTE het-erogeneous networksrdquo in Proceedings of the IEEE InternationalSymposium on Wireless Personal Multimedia Communications(WPMC rsquo14) pp 75ndash79 September 2014

[23] H Pervaiz L Musavian and Q Ni ldquoJoint user associationand energy-efficient resource allocation with minimum-rate

constraints in two-tier HetNetsrdquo in Proceedings of the IEEE24th Annual International Symposium on Personal Indoor andMobile Radio Communications (PIMRC rsquo13) pp 1634ndash1639London UK September 2013

[24] V N Ha and L B Le ldquoDistributed base station associationand power control for heterogeneous cellular networksrdquo IEEETransactions onVehicular Technology vol 63 no 1 pp 282ndash2962014

[25] A M Rao A Weber S Gollamudi and R Soni ldquoLTE andHSPA+ revolutionary and evolutionary solutions for globalmobile broadbandrdquo Bell Labs Technical Journal vol 13 no 4pp 7ndash34 2009

[26] S S Mwanje and A Mitschele-Thiel ldquoMinimizing handoverperformance degradation due to LTE self organized mobilityload balancingrdquo in Proceedings of the IEEE 77th VehicularTechnology Conference (VTC Spring rsquo13) June 2013

[27] 3GPP TR 36942 Ver 1100 Rel 11 ldquoTechnical SpecificationGroup Radio Access Network Evolved Universal TerrestrialRadio Access (EUTRA) Radio Frequency (RF) system scenar-iosrdquo Tech Rep 2012

[28] R Combes Z Altman and E Altman ldquoScheduling gain forfrequency-selective Rayleigh-fading channels with applicationto self-organizing packet schedulingrdquo Performance Evaluationvol 68 no 8 pp 690ndash709 2011

[29] R Combes S E Elayoubi and Z Altman ldquoCross-layer anal-ysis of scheduling gains application to LMMSE receivers infrequency-selective Rayleigh-fading channelsrdquo inProceedings ofthe IEEE International Symposium on Modeling and Optimiza-tion in Mobile Ad Hoc and Wireless Networks (WiOpt rsquo11) pp133ndash139 Princeton NJ USA May 2011

[30] H J Kushner and P AWhiting ldquoConvergence of proportional-fair sharing algorithms under general conditionsrdquo IEEE Trans-actions onWireless Communications vol 3 no 4 pp 1250ndash12592004

[31] C Tellambura ldquoComputing the outage probability in mobileradio networks using the sampling theoremrdquo IEEE Transactionson Communications vol 47 no 8 pp 1125ndash1128 1999

[32] S Tang B L Mark and A E Leu ldquoAn exact solution for outageprobability in cellular networksrdquo in Proceedings of the MilitaryCommunications Conference (MILCOM rsquo07) October 2007

[33] H Boostanimehr and V K Bhargava ldquoDistributed and QoS-driven cell association in HetNets to minimize global outageprobabilityrdquo in Proceedings of the IEEE Global CommunicationsConference (GLOBECOM rsquo14) pp 3665ndash3671 IEEE AustinTex USA December 2014

Submit your manuscripts athttpswwwhindawicom

Computer Games Technology

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Distributed Sensor Networks


Advances in

FuzzySystems

Hindawi Publishing Corporationhttpwwwhindawicom

Volume 2014


ReconfigurableComputing

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014


Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

Artificial Intelligence

HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014

Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi Publishing Corporation

httpwwwhindawicom Volume 2014

Advances in

Multimedia


Biomedical Imaging


ArtificialNeural Systems

Advances in


RoboticsJournal of



Computational Intelligence and Neuroscience

Industrial EngineeringJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Human-ComputerInteraction

Advances in

Computer EngineeringAdvances in



a use association policy is proposed to minimize the energyconsumption in LTE access networks which are composedof small cells deployed densely However cell load-awareassociationmethods require that cells broadcast the cell loadsperiodically or on-demand In addition MSs select a cell toassociate with to maximize its benefit selfishly while cellspassively accept the association request fromMSsThereforesystem resources are not used in an optimal way because ofthe selfishness

To address this issue we devise a distributed cell-guidedassociation method (CGAM) In CGAM cells do not pas-sively accept the association requests from MSs On the con-trary cells determine whether or not to accept the associationrequest based on their resource availabilityThus even if MSsoperate selfishly to maximize its payoff system can guideMSs to achieve its own objective such as increasing resourceutilization and cell load balance In addition since CGAMdoes not require cells to broadcasts cell load conventional celloperations need not be changed

The rest of the paper is organized as follows In Section 2we review related works on association management InSection 3 we describe the CGAM in detail after introducingsystemmodel In Section 4we present anddiscuss simulationresults to verify CGAM by comparing the performances ofCGAM and SINR-based association method and cell load-aware association method

2 Related Works

Since the cell association problem is to find a set of MSsand cells pairs that optimize a given objective function ofa wireless network optimization methods are widely usedto design an association rule Various objective functionsare devised for the cell association problem In [7] authorspropose an algorithm that maximizes the system revenuewhile associating MSs with the minimum total transmissionpower They use Bendersrsquo decomposition to solve nonconvexoptimization problem optimally Throughput maximizationis also used as an objective function In [8] sum rate ismaximized and authors in [9 10] maximize the log-utility ofa network under the proportional fairness In [11] MSs areassociated with BSs in a way that global outage probability isminimizedHowever it has been shown that the optimizationproblem is an NP-hard problem

To overcome the NP-hardness different kinds of opti-mization techniques are used to solve the cell associationproblem Markov decision process is used in [12 13] How-ever it is hard to define state transition probability andthe complexity increases substantially as the number ofstates increases In [14] the structure of the network utilitymaximization problem is used to solve the problem directlyDual decomposition is often employed to design a distributedalgorithm by relaxing the optimization constraints [10 15]However since the computational cost of these methods ishigh it is not clear whether these methods can be applied ata small time scale when the values of input variables changevery fast

Game theory is also applied to solve the cell associationproblem In [16] an evolutionary game theory is used

to devise an algorithm for cell association and antennaallocation in 5G networks with massive MIMO Authorsin [17] show that the game that users selfishly select basestations giving them the best throughput and BSs allocatethe same time to their users has one Nash equilibrium pointwhich achieves proportional fairness system-wide Howeverthe convergence of the algorithms based on a game theoryis not guaranteed generally Additionally in terms of theimplementation they usually cause large overhead whichdeteriorates resource utilization

Stochastic geometry has been used to analyze the cellassociation problem In [18] authors propose a distributedbelief propagation algorithm to resolve user association prob-lem and analyze the average sparsity and degree distributionusing stochastic geometry Stochastic geometry is also usedin [19] to analyze the service success probability Then theyderive the impact of cell association and user schedulingon the service success probability Unlike the optimizationmethods that maximize the utility function for the currentnetwork configuration stochastic geometry performs opti-mization over the average utility Therefore even though thecomplexity and overhead of a stochastic geometry approachare lower than those of repeating optimization processwhenever network configuration changes the results will besuboptimal

Unlike the association methods based on a theoreticalstatic interference model a practical measurement basedinterference-aware association policy is proposed in [20]

There have been research works that optimize jointlycell association and other performance metrics In [21] atractable framework is proposed to analyze the performanceof eICIC by jointly considering cell association resourcepartitioning and transmit power reduction An energy effi-cient user association scheme is proposed in [22 23] JointBS association and power control algorithm that updatesiteratively BS association solution then the transmit power ofeach user is proposed in [24]

3 Cell-Guided Association Method

31 System Model We consider a downlink of a wireless net-work that provides best effort data service to MSs We denoteby119873119888 the set of cells in a network and by |119873119888| the cardinalityof119873119888 We assume that the system fully reuses frequency

Since radio resources of a cell are allocated to MSs inthe unit of radio resource block (RB) in modern wirelessnetworks such as LTE we assume that each cell 119909 has 119877119909 RBsWe consider long term SINR and throughput In other wordswe assume that the timescale measuring these values is muchlarger than that of fast channel variation Thus if we denotethe transmit power of a cell 119909 as 119875119909 the SINR between a MS119910 and a cell 119909 is given by

119878119909119910 =119875119909119866119909119910

119860119873 + sum119888isin119873119888minus119909 119875119888119866119888119910 (1)

where 119860119873 is the noise power and 119866119909119910 denotes the channelgain between a cell 119909 and a MS 119910 which includes the system




119863119909119910 = 119877119909119910119882119877log2 (1 + 119878119909119910) (2)



119877119909119910 =1198771199091003816100381610038161003816119872119909

1003816100381610038161003816119892 (119872119909) (3)



119874119909119910 = Prob (119863119909119910 lt 120579119889)

= 1

minus 119890

119860119873119861119909119910119866119909119910119875119909 prod

119894isin119873119888minus119909

1 (119866119894119910119875119894)

119861119909119910 (119866119909119910119875119909) + 1 (119866119894119910119875119894)

(4)

























0

01

02

03

04

05

06

07

08

09

1

0 500 1000 1500 2000

Prop

ortio

n of

MSs


(a) All MSs

0

01

02

03

04

05

06

07

08

09

1

0 500 1000 1500 2000

Prop

ortio

n of

MSs


(b) Edge MSs


0

01

02

03

04

05

06

07

08

09

1

0 005 01 015 02 025 03 035 04 045 05

Prop

ortio

n of

MSs


(a) All MSs

0

01

02

03

04

05

06

07

08

09

1

0 01 02 03 04 05 06 07

Prop

ortio

n of

MSs


SNMONMSOM

(b) Edge MSs






708090

100110120130140

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM


0005

01015

02025

03035

04045

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM


0010203040506070809

1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Out

age p

rob

Cell IDSNMONMSOM


5 Conclusions



Competing Interests


Acknowledgments


References












































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






119863119909119910 = 119877119909119910119882119877log2 (1 + 119878119909119910) (2)



119877119909119910 =1198771199091003816100381610038161003816119872119909

1003816100381610038161003816119892 (119872119909) (3)



119874119909119910 = Prob (119863119909119910 lt 120579119889)

= 1

minus 119890

119860119873119861119909119910119866119909119910119875119909 prod

119894isin119873119888minus119909

1 (119866119894119910119875119894)

119861119909119910 (119866119909119910119875119909) + 1 (119866119894119910119875119894)

(4)

























0

01

02

03

04

05

06

07

08

09

1

0 500 1000 1500 2000

Prop

ortio

n of

MSs


(a) All MSs

0

01

02

03

04

05

06

07

08

09

1

0 500 1000 1500 2000

Prop

ortio

n of

MSs


(b) Edge MSs


0

01

02

03

04

05

06

07

08

09

1

0 005 01 015 02 025 03 035 04 045 05

Prop

ortio

n of

MSs


(a) All MSs

0

01

02

03

04

05

06

07

08

09

1

0 01 02 03 04 05 06 07

Prop

ortio

n of

MSs


SNMONMSOM

(b) Edge MSs






708090

100110120130140

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM


0005

01015

02025

03035

04045

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM


0010203040506070809

1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Out

age p

rob

Cell IDSNMONMSOM


5 Conclusions



Competing Interests


Acknowledgments


References












































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in

















0

01

02

03

04

05

06

07

08

09

1

0 500 1000 1500 2000

Prop

ortio

n of

MSs


(a) All MSs

0

01

02

03

04

05

06

07

08

09

1

0 500 1000 1500 2000

Prop

ortio

n of

MSs


(b) Edge MSs


0

01

02

03

04

05

06

07

08

09

1

0 005 01 015 02 025 03 035 04 045 05

Prop

ortio

n of

MSs


(a) All MSs

0

01

02

03

04

05

06

07

08

09

1

0 01 02 03 04 05 06 07

Prop

ortio

n of

MSs


SNMONMSOM

(b) Edge MSs






708090

100110120130140

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM


0005

01015

02025

03035

04045

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM


0010203040506070809

1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Out

age p

rob

Cell IDSNMONMSOM


5 Conclusions



Competing Interests


Acknowledgments


References












































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




708090

100110120130140

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM


0005

01015

02025

03035

04045

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Dat

a rat

e (M

bps)

Cell IDSNMONMSOM


0010203040506070809

1

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Out

age p

rob

Cell IDSNMONMSOM


5 Conclusions



Competing Interests


Acknowledgments


References












































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in



Distributed Association Method Assisted by Cell for...

Documents

Transcript of Distributed Association Method Assisted by Cell for...