Internet Protocol Version 6

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 3, MARCH 2013 1077

Comparative Handover Performance Analysis ofIPv6 Mobility Management Protocols

Jong-Hyouk Lee, Member, IEEE, Jean-Marie Bonnin, Senior Member, IEEE, Ilsun You, andTai-Myoung Chung, Senior Member, IEEE

Abstract—IPv6 mobility management is one of the most chal-lenging research topics for enabling mobility service in the forth-coming mobile wireless ecosystems. The Internet EngineeringTask Force has been working for developing efficient IPv6 mobilitymanagement protocols. As a result, Mobile IPv6 and its extensionssuch as Fast Mobile IPv6 and Hierarchical Mobile IPv6 have beendeveloped as host-based mobility management protocols. Whilethe host-based mobility management protocols were being en-hanced, the network-based mobility management protocols suchas Proxy Mobile IPv6 (PMIPv6) and Fast Proxy Mobile IPv6(FPMIPv6) have been standardized. In this paper, we analyze andcompare existing IPv6 mobility management protocols includingthe recently standardized PMIPv6 and FPMIPv6. We identifyeach IPv6 mobility management protocol’s characteristics andperformance indicators by examining handover operations. Then,we analyze the performance of the IPv6 mobility managementprotocols in terms of handover latency, handover blocking prob-ability, and packet loss. Through the conducted numerical results,we summarize considerations for handover performance.

Index Terms—Fast Mobile IPv6 (FMIPv6), Fast Proxy MobileIPv6 (FPMIPv6), Hierarchical Mobile IPv6 (HMIPv6), MobileIPv6 (MIPv6), Proxy Mobile IPv6 (PMIPv6).

I. INTRODUCTION

MOBILE wireless ecosystems facilitate more rapidgrowth of digital ecosystems for our human lives

[1]–[6]. Mobility management protocols are at the heart of themobile wireless ecosystems. Mobile social networking, mobilecollaboration computing, and mobile shopping shall become areality with a well-deployed mobility management architecture.

Various mobility management protocols for enabling mo-bility service have been introduced. In particular, mobilitysupport in the network layer has been being developed by theInternet Engineering Task Force (IETF). Since the Mobile IPv6(MIPv6) specification [7] was published, extensions includingFast Mobile IPv6 (FMIPv6) [8] and Hierarchical Mobile IPv6(HMIPv6) [9] for enhancing the performance of MIPv6 havebeen developed. During the time when the extensions to MIPv6

Manuscript received August 23, 2011; revised March 5, 2012; acceptedApril 18, 2012. Date of publication May 4, 2012; date of current versionOctober 16, 2012.

J.-H. Lee and J.-M. Bonnin are with the Networks, Security and Multimedia(RSM) Department, TELECOM Bretagne, 35576 Cesson-Sévigné, France(e-mail: [email protected]; [email protected]).

I. You is with the School of Information Science, Korean Bible University,Seoul 139-791, Korea (e-mail: [email protected]).

T.-M. Chung is with Sungkyunkwan University, Suwon 440-746, Korea(e-mail: [email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TIE.2012.2198035

were developed, comparative performance analysis for IPv6mobility management protocols has been used as inputs fordeveloping improvements [10], [11]. For instance, comparativeperformance analysis studied for MIPv6, FMIPv6, HMIPv6,and a combination of FMIPv6 and HMIPv6 has been carriedout in [12] and [13] that identify each mobility managementprotocol’s characteristics and performance indicators.

While host-based mobility management protocols are de-ployable in wireless mobile communication infrastructures,communication service providers and standards developmentorganizations have recognized that such conventional solutionsfor mobility service are not suitable; in particular, for telecom-munication service, a mobile node (MN) is required to havemobility functionalities at its network protocol stack inside,and thus, modifications or upgrades of the MN are forced. Itobviously increases the operation expense and complexity forthe MN. The host-based mobility management protocols alsocause lack of control for operators since the MN manages itsown mobility support. Accordingly, a new approach to supportmobility service has been required and pushed by the 3rdGeneration Partnership Project to the IETF.

Proxy Mobile IPv6 (PMIPv6) is a network-based mobilitymanagement protocol that allows an MN to change its pointof attachment without any mobility signaling processed at theMN [14]. Two types of mobility service provisioning entity areintroduced in PMIPv6: mobility access gateway (MAG) andlocal mobility anchor (LMA). A MAG is a mobility serviceprovisioning entity which is responsible for detecting and reg-istering the movement of the MN in its access network. Asthe MAG detects the movement of the MN, it sends a proxybinding update (BU) (PBU) message to the LMA. Note that theLMA operates as a home agent (HA) as specified in [7] and alsoinvolves additional functions. As it receives the PBU messagefor the MN, the LMA recognizes that the MN has attached tothe MAG and creates/updates the binding cache for the MN.The MAG receives the proxy binding acknowledgment (BAck)(PBAck) message including the home network prefix (HNP) forthe MN and then sends the router advertisement (RA) messageincluding the HNP. The MN configures its address, proxy homeaddress (pHoA), based on the HNP included in the RA messagesent from the MAG in the access network. Because the LMAalways provisions the same HNP for a given MN during itsmovements, the MN obtains the same pHoA within the PMIPv6domain. Owing to the network-based mobility service providedby mobility service provisioning entities, the entire PMIPv6domain appears as a single link from the perspective of the MN[14]. As an extension protocol to PMIPv6, Fast Proxy Mobile

0278-0046/$31.00 © 2012 IEEE

1078 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 3, MARCH 2013

IPv6 (FPMIPv6) [15] has been later developed to acceleratethe handover performance by reducing handover latency andpreventing packet loss.

Compared to host-based mobility management protocolswhich have been evaluated over the years, the network-basedmobility management protocols such as PMIPv6 and FPMIPv6are in the early stage for deployments. It is thus desirable toanalyze and compare the host-based mobility management pro-tocols and the network-based mobility management protocolstogether. Note that the reader is assumed to be familiar with thedetails of MIPv6, FMIPv6, HMIPv6, PMIPv6, and FPMIPv6because this paper directly goes through the analytical model-ing and performance evaluation of those protocols.

In this paper, we report a performance evaluation analysis. Inparticular, to the best of our knowledge, such numerical perfor-mance analysis, including MIPv6, FMIPv6, HMIPv6, PMIPv6,and FPMIPv6, is unprecedented in the literature. The analysisconducted in this paper also provides a list of considerations forhandover performance:

1) utilizing link-layer (L2) information that helps to preparean MN’s handover before the MN attaches to a newaccess network;

2) employing buffering management that helps to preventpacket loss during the MN’s handover;

3) wireless link condition that largely affects the handoverperformance of all mobility management protocols;

4) address configuration and preparation that count for alarge portion of handover latency of host-based mobilitymanagement protocols;

5) network topology that affects the handover performanceof all mobility management protocols.

The remainder of this paper is organized as follows. InSection II, previous works for performance analysis of IPv6mobility management protocols are reviewed. Then, as prelim-inaries, the performance metrics, considered network model,message information, and packet transportation delay modelsare presented in Section III. In Section IV, the analyticalmodeling for performance evaluation is presented. The com-prehensive numerical analysis and discussions are presented inSection V. Finally, conclusions are given in Section VI.

II. LITERATURE REVIEW

In this section, we present some of previous studies forperformance analysis of IPv6 mobility management protocols.

In [12], the authors have carried out a performance compar-ison among MIPv6, FMIPv6, HMIPv6, and a combination ofFMIPv6 and HMIPv6. Simulation using the network simulatorns-2 has been performed to analyze signaling costs associ-ated to the different IPv6 mobility management protocols.The authors showed that the protocol combining of FMIPv6and HMIPv6 outperforms the other protocols in most cases.However, the combination of FMIPv6 and HMIPv6 resultedin a worse performance than MIPv6 when a user packet rateis low.

In [13], the authors have developed an analytical frameworkfor performance analysis of IPv6 mobility management proto-cols. MIPv6, FMIPv6, HMIPv6, and a combination of FMIPv6

and HMIPv6 have been compared and evaluated in terms of sig-naling cost, binding refresh cost, packet delivery cost, requiredbuffer space, and handover latency. In the paper, the authorspresented the effect of subnet residence time, packet arrivalrate, and wireless link delay to the different IPv6 mobilitymanagement protocols.

Simple handover performance analysis has been presentedin [16]. In the paper, the authors showed that PMIPv6 out-performs other IPv6 mobility management protocols owing toits simple handover procedure. In [17], HMIPv6 and PMIPv6are compared and analyzed in terms of location update, packetdelivery, and wireless power consumption costs. Then, in [18],four different route optimization (RO) schemes for PMIPv6are presented and analyzed. In the paper, the authors haveshowed that the router optimization schemes solve the ineffec-tive routing path problem and argued that the scalability of thePMIPv6 architecture is improved owing to distributed routingpaths in the router optimization schemes. In [19], an analyticalcost model has been developed for evaluating the performanceof IPv6 mobility management protocols. The IPv6 mobilitymanagement protocols such as MIPv6, FMIPv6, HMIPv6, andPMIPv6 are analyzed and compared in terms of signaling cost,packet delivery cost, tunneling cost, and total cost.

However, the previous performance analysis studies [12],[13] considered only the host-based mobility management pro-tocols. In [16] and [19], PMIPv6 has been compared with thehost-based mobility management protocols, but the recentlydeveloped FPMIPv6 protocol [15] has not been considered.Moreover, the cost analysis studies performed in [13] and[17]–[19] do not help to understand the handover performanceof IPv6 mobility management protocols.

In this paper, we develop a uniform framework for conduct-ing analytic modeling across the spectrum of IPv6 mobilitymanagement protocols. The host-based mobility manage-ment protocols such as MIPv6, FMIPv6, and HMIPv6 andthe network-based mobility management protocols such asPMIPv6 and FPMIPv6 are analyzed and compared in termsof handover latency, handover blocking probability, andpacket loss.

III. PRELIMINARIES

A. Performance Metrics

The following performance metrics are used.1) Handover latency: It is the time interval during which an

MN cannot send or receive any packets while it performsits handover between different access networks.

2) Handover blocking probability: It is the probability whichan MN cannot complete its handover when the networkresidence time is less than the handover latency.

3) Packet loss: It is the sum of all lost packets destined foran MN during the MN’s handover.

B. Considered Network Model

The considered network model is depicted in Fig. 1 show-ing a generic network topology wherein all communicationentities are displayed. Suppose that the MN changes its point

LEE et al.: COMPARATIVE HANDOVER PERFORMANCE ANALYSIS OF IPv6 MOBILITY MANAGEMENT PROTOCOLS 1079

Fig. 1. Considered network model.

of attachments in a given domain composed of several accessrouters (ARs). That is, the movement of the MN is limitedin the domain where the gate located at the top level of thedomain acts as an edge router connected to the Internet. Underthe assumption, the gate can be treated as the mobility anchorpoint (MAP) for HMIPv6 or the LMA for PMIPv6. Similarly,the MAG can be located at the AR when PMIPv6 is consideredin the network model shown in Fig. 1.

In Fig. 1, the following hop count parameters are defined fordescribing particular paths between communication entities.

1) hC−H : It is the average number of hopsbetween the correspondent node (CN) and the HA.

2) hC−G: It is the average number of hops between the CNand the gate.

3) hH−G: It is the average number of hops between the HAand the gate.

4) hG−A: It is the average number of hops between the gateand the AR.

5) hA−A: It is the average number of hops between theneighbor ARs.

6) hA−M : It is the average number of hops between theAR and the MN. Since hA−M is the wireless link, it isassumed to be one.

According to the considered network model, data/controlpackets being exchanged between the MN and the HA/CNmust be routed through the gate. For instance, when RO inMIPv6 is enabled, data packets sent from the CN to the MNtravel through hC−G + hG−A + hA−M , where hA−M is thewireless link established between the MN and the serving AR.In addition, hA−A can be rewritten as

√hG−A [20], [21].

C. Messages Related to Mobility Support

Various messages related to mobility support are used in IPv6mobility management protocols. The following message sizesin bytes are considered in our analytical modeling [22], [18].

1) LRS: It is the size of the router solicitation (RS) message,which is 52.

2) LRA: It is the size of the RA message, which is 80.

3) LBU−HA: It is the size of the BU message sent from theMN to the HA, which is 56.

4) LBAck−HA: It is the size of the BAck message, whichis 56.

5) LBU−CN: It is the size of the BU message sent from theMN to the CN, which is 66.

6) LLBU−MAP: It is the size of the local BU (LBU) messagesent from the MN to the MAP, which is 56.

7) LLBAck−MAP: It is the size of the local BAck (LBAck)message, which is 56.

8) LPBU−LMA: It is the size of the PBU message sent fromthe MAG to the LMA, which is 76.

9) LPBAck−LMA: It is the size of the PBAck message, whichis 76.

10) LHoTI: It is the size of the home test (HoT) init (HoTI)message, which is 64.

11) LCoTI: It is the size of the care-of test (CoT) init (CoTI)message, which is 64.

12) LHoT: It is the size of the HoT message, which is 74.13) LCoT: It is the size of the CoT message, which is 74.14) LFBU: It is the size of the fast BU (FBU) message, which

is 56.15) LFBAck: It is the size of the fast BAck (FBAck) message,

which is 56.16) LUNA: It is the size of the unsolicited neighbor advertise-

ment (NA) (UNA) message, which is 52.17) LRtSolPr: It is the size of the RS for proxy advertisement

(RtSolPr) message, which is 52.18) LPrRtAdv: It is the size of the proxy RA (PrRtAdv)

message, which is 80.19) LHI: It is the size of the handover initiate (HI) message,

which is 52.20) LHAck: It is the size of the handover acknowledge (HAck)

message, which is 52.21) LT : It is the size of the tunneling header, which is 40.22) LD: It is the size of the user data packet, which is 120.

D. One-Way Packet Transportation Delay Over aWireless Link

Wireless links are unreliable particularly compared to wiredlinks. An MN is attached to its AR through a wireless link,and data/control packets for the MN are transmitted over thewireless link. Accordingly, the packet transportation delay overthe wireless link is a critical performance factor. The reportedresults in [23]–[25] are used here. Suppose that τ and ρf denotethe interframe time and the frame error rate (FER) over thewireless link, respectively. Let pi,j be the probability that thefirst frame sent from the MN arrived at the AR successfully,being the ith retransmitted frame at the jth retransmission trial.Then, the one-way frame transportation delay dframe betweenthe MN and the AR through the wireless link is expressed asfollows [23]–[25]:

dframe = Dwl(1− ρf ) +

n∑i=1

i∑j=1

pi,j (2i×Dwl + 2(j − 1)τ)

(1)


Fig. 2. Timing diagram for MIPv6 handover.

where i ≤ n, j ≤ i, and Dwl is the wireless link delay mainlydepending on which L2 technology is being used. In addition,n is the maximum number of retransmission trials. Then, pi,jin (1) is expressed as follows [23]–[25]:

pi,j = ρf (1− ρf )2 ((2− ρf )ρf )

((i2−i)/2)+j−1) . (2)

Suppose that k denotes the number of frames per packet overthe wireless link. Then, k is expressed as follows:

k =

⌈Lp

Lf

⌉(3)

where Lp and Lf are the packet size and the frame size,respectively. Thus, by combining (1)–(3), the one-way packettransportation delay over the wireless link dwl(Lp) is obtainedas follows:

dwl(Lp) = dframe + (k − 1)τ. (4)

E. One-Way Packet Transportation Delay Over a Wired Link

Wired links are reliable compared to wireless links. Assum-ing that packets sent over wired links will not be lost and willreach the destination without retransmission trials, the one-way packet transportation delay over the wired link dwd(Lp)is simply obtained as follows:

dwd(Lp) =Lp

BWwired+Dwired (5)

where BWwired and Dwired are the bandwidth and the latencyof wired links, respectively. Then, by considering the numberof hops between the two end nodes, the one-way packet trans-portation delay over a number of wired links dwd(Lp, h) isobtained as follows [26]:

dwd(Lp, h) =Lp × h

BWwired+Dwired (6)

where h is the number of hops from the source node to thedestination node.

IV. ANALYTICAL MODELING OF IPv6 MOBILITY

MANAGEMENT PROTOCOLS

In this section, formulas are derived for analyzing the perfor-mance metrics based on the handover timing diagrams.

A. Handover Latency of MIPv6

Fig. 2 shows the timing diagram for MIPv6 handover. Theactual handover of an MN is started when the MN losesconnectivity. Then, the MN attaches to an access network asits link goes up and performs the movement detection processby sending the RS message in order to receive the RA messagequickly. The MN configures its new care-of address (CoA)based on the network prefix information included in the RAmessage, and it performs the duplicate address detection (DAD)process. Note that the stateless address autoconfiguration isassumed here. In the case that the CoA is valid to be used in thenew network, the MN registers its new location information bysending BU messages to its HA and CN. For the CN, the HoTIand CoTI messages are sent to start the return routability (RR)process.1 The HoTI message is first tunneled to the HA, andthen, it is forwarded to the CN. Receiving the BAck messagesent from the HA, it indicates that the location update to theHA is completed, whereas the actual location update to the CNis started by sending the BU message to the CN after receivingthe valid HoT and CoT messages. When the CN receives theBU message sent from the MN, it starts to send data packetsdirectly to the MN. Note that the CN does not need to send theBAck message back to the MN [7].

Suppose that L(MIPv6)HO is the handover latency of MIPv6.

Then, it is expressed as follows:

L(MIPv6)HO = TL2 + TMD + TDAD + TR (7)

where TL2 is the L2 handover latency, TMD is the movementdetection latency, TDAD is the DAD latency, and TR is theregistration latency. TL2 depends on which L2 technology andmanufacture chipset are being used. The movement detectionprocess is completed as the MN receives the solicited RAmessage from the AR in the new access network. Accordingly,if the movement detection process is immediately started withthe linkup signaling, TMD can be rewritten as

TMD = dwl(LRS) + dwl(LRA). (8)

In Fig. 2, TH−M presents the location update time forthe HA. Suppose that hH−A is the average number of hops

1As described in [17, Sec. 11.6.1], in some cases, the RR process maybe completed with only one message pair exchange or even be completedwithout any message exchange. However, in this paper, we assume that theMN performs its RR process for each handover.


Fig. 3. Timing diagram for predictive FMIPv6 handover.

Fig. 4. Timing diagram for reactive FMIPv6 handover.

between the HA and the AR serving the MN. Then, TH−M isexpressed as

TH−M = dwl(LBU−HA) + dwd(LBU−HA, hH−A)

+ dwl(LBAck−HA) + dwd(LBAck−HA, hH−A). (9)

The DAD process is successfully completed if a defendingNA message for the generated CoA is not arrived in Re-transTimer [27]. Accordingly, TDAD can be rewritten as thevalue of RetransTimer defined in [28]. When RO in MIPv6is enabled, the RR process must be performed, so TR can berewritten as

TR = max{Tα, Tβ}+ TC−M (10)

where Tα is the required time to exchange the HoTI and HoTmessages via the HA and Tβ is the required time to exchangethe CoTI and CoT messages directly with the CN. Suppose thathC−A is the average number of hops between the CN and theAR serving the MN, i.e., hC−A = hC−G + hG−A. Then, Tα

and Tβ can be expressed as

Tα = dwl(LHoTI) + dwd(LHoTI, hH−A + hC−H)

+ dwl(LHoT) + dwd(LHoT, hH−A + hC−H) (11)

Tβ = dwl(LCoTI) + dwd(LCoTI, hC−A)

+ dwl(LCoT) + dwd(LCoT, hC−A). (12)

In (10), TC−M is the time required to send the BU messageto the CN and receive the first data packet sent from the CN.Note that the CN’s BAck message is not required as presentedin [7]. Then, TC−M is expressed as

TC−M = dwl(LBU−CN) + dwd(LBU−CN, hC−A)

+ dwl(LD) + dwd(LD, hC−A). (13)

B. Handover Latency of FMIPv6

FMIPv6 operates in either the predictive mode or the reactivemode, depending on the circumstances [10].

Fig. 3 shows the timing diagram for predictive FMIPv6handover. By utilizing the L2 trigger, an MN anticipates itsmovement to reduce its handover latency and also preventpacket loss. Predictive FMIPv6 is performed when the MN suc-cessfully receives the FBAck message sent from the previousAR (pAR) before it moves to the new AR (nAR). As shownin Fig. 3, the MN prepares its handover at the previous accessnetwork. For instance, the MN actively obtains the new CoA(NCoA) which will be used in the new access network, whereasthe relevant ARs exchange required information for serving theMN. Then, as the MN attaches to the new access network, itimmediately sends the UNA message with the NCoA alreadygenerated while being attached at the previous access network.Thus, the handover latency in predictive FMIPv6 is signifi-cantly reduced compared to that of MIPv6.

Suppose that L(Pre-FMIPv6)HO is the handover latency of pre-

dictive FMIPv6. Then, it is expressed as follows:

L(Pre-FMIPv6)HO = TL2 + TPRE (14)

where TPRE represents the time at which the nAR receives theUNA message sent from the MN and the time at which the MNreceives the first data packet sent from the nAR. Note that thedata packets sent to the MN are the buffered data packets thatthe pAR has forwarded. Then, TPRE is expressed as follows:

TPRE = dwl(LUNA) + dwl(LD). (15)

Fig. 4 shows the timing diagram for reactive FMIPv6 hand-over. Even if an MN can anticipate its movement by utilizingthe L2 trigger, sometime, the MN cannot complete its handoverpreparing at the previous access network. That is, reactiveFMIPv6 handover is performed when the MN cannot receivethe FBAck message sent from the pAR [10].


Fig. 5. Timing diagram for HMIPv6 handover.

Fig. 6. Timing diagram for PMIPv6 handover.

Suppose that L(Re-FMIPv6)HO is the handover latency of reac-

tive FMIPv6. Then, it is expressed as follows:

L(Re-FMIPv6)HO = TL2 + TDAD + TRE (16)

where TDAD is included due to the lack of handover preparingat the previous access network. Similarly, TRE is included forrepresenting the times to send the FBU message, exchangerequired information between the relevant ARs, and receive thefirst data packet sent from the nAR. TRE is expressed as

TRE = dwl(LFBU) + dwd(LFBU, hA−A)

+ dwd(LHI, hA−A) + dbuff-packet (17)

where dbuff-packet is the time which the first data packetbuffered at the pAR arrives at the MN via the nAR. The buffereddata packets at the pAR are immediately sent to the nAR withthe FBAck message. Accordingly, dbuff-packet is expressed as

dbuff-packet = dwd(LD + LT , hA−A) + dwl(LD) (18)

where LT is considered because the pAR tunnels data packetsdestined for the MN to the nAR.

C. Handover Latency of HMIPv6

Fig. 5 shows the timing diagram for HMIPv6 handover.HMIPv6 manages the movement of an MN in a localizedmanner. In the diagram, the movements of the MN are assumedas intradomain handovers. That is, the MN changes its point ofattachment within a MAP domain. In HMIPv6, L2 informationis not utilized to anticipate the movement of the MN so thatthe handover process of HMIPv6 is similar to that of MIPv6.The MN only registers its new location information by sendingthe LBU message with the LCoA to its MAP. The actions forregistering new location information to both of the HA andCN are not required in HMIPv6. This is because the MN’smovement within the MAP domain is transparent to the outsideof the MAP domain [11], [16].

Suppose that L(HMIPv6)HO is the handover latency of HMIPv6.


L(HMIPv6)HO = TL2 + TMD + TDAD + TMAP (19)

where TMD and TDAD are included. This is because HMIPv6does not utilize L2 information to improve handover speed andthe LCoA is required to be generated as the MN receives theRA message at the new access network. Then, TMAP representsthe required time to send the LBU message, receive the LBAckmessage, and also receive the data packet sent from the MAP.Then, TMAP is expressed as follows:

TMAP = dwl(LBU−MAP) + dwd(LBU−MAP, hG−A)

+ dmap-packet (20)

where dmap-packet is the time which the first data packet sentfrom the MAP arrives at the MN. The MAP immediately sendsdata packets destined for the MN with the LBAck message.Accordingly, dmap-packet is expressed as

dmap-packet = dwl(LD + LT ) + dwd(LD + LT , hG−A) (21)

where LT is taken into account because the data packets sentfrom the MAP to the MN are tunneled.

D. Handover Latency of PMIPv6

Fig. 6 shows the timing diagram for PMIPv6 handover.Similar to HMIPv6, PMIPv6 manages the movement of anMN in a localized manner as well, but mobility service forthe MN is supported by mobility service provisioning entities[17], [18]. As the MN attaches to the new access network, itsmovement is detected and registered by the MAG at the newaccess network. Then, the MN obtains the same HNP includedin the RA message sent from the MAG at the new accessnetwork so that the address configuration and DAD process arenot required when the MN performs its handover in a PMIPv6domain [16].


Fig. 7. Timing diagram for predictive FPMIPv6 handover.

Fig. 8. Timing diagram for reactive FPMIPv6 handover.

Suppose that L(PMIPv6)HO is the handover latency of PMIPv6.


L(PMIPv6)HO = TL2 + TLMA (22)

where TLMA involves the required time to send the RS message,exchange PBU/PBAck messages between the MAG and theLMA, and receive the first data packet sent from the LMA. Inthis paper, we assume that the MAG detects the movement ofthe MN when the MAG receives the RS message sent from theMN. Then, TLMA is expressed as follows:

TLMA = dwl(LRS) + dwd(LPBU, hG−A) + dlma-packet (23)

where dlma-packet is the time which the first data packet sentfrom the LMA arrives at the MN. A bidirectional tunnelbetween the LMA and the MAG can be implemented as astatic tunneling between them that requires no additional tun-neling establishment latency. Here, such a static tunneling isconsidered for PMIPv6. As the LMA receives the valid PBUmessage sent from the MAG, it sends data packets destined forthe MN with the PBAck message. Accordingly, dlma-packet isexpressed as

dlma-packet = dwl(LD) + dwd(LD + LT , hG−A) (24)

where LT is only taken into account at dwd(Lp, h). This is be-cause the data packets for the MN are only tunneled between theLMA and the MAG. Notice that this is a difference compared tothat of HMIPv6. Even if both of PMIPv6 and HMIPv6 similarlymanage the MN in a localized manner, PMIPv6 further reducesthe packet transportation overhead over the wireless link [17].

E. Handover Latency of FPMIPv6

Similar to FMIPv6, FPMIPv6 consists of predictive andreactive modes.

Fig. 7 shows the timing diagram for predictive FPMIPv6 han-dover. While an MN is attached to a previous MAG (pMAG),it reports an imminent handover event to the pMAG. Pre-

dictive FPMIPv6 is performed when the pMAG successfullyexchanges the required information of the MN with a new MAG(nMAG) via the HI and HAck messages before the MN attachesto the nMAG. After a successful HI/HAck message exchange,the bidirectional tunnel between the pMAG and nMAG isestablished. The pMAG uses this tunnel to forward data packetsdestined for the MN to the nMAG. When the MN changes itspoint of attachment to the nMAG, the forwarded data packetswill be directly sent to the MN from the nMAG.

Suppose that L(Pre-FPMIPv6)HO is the handover latency of

predictive FPMIPv6. Then, it is expressed as follows:

L(Pre-FPMIPv6)HO = TL2 + TPRE-P (25)

where TPRE-P is composed of the sum of the IP-layer con-nection setup delay Dπ and the first data packet arrival delayfrom the nMAG to the MN dmag-packet. Accordingly, TPRE-Pis expressed as follows:

TPRE-P = Dπ + dmag-packet (26)

where Dπ is assumed to be the same delay as dwl(LUNA) inthis paper and dmag-packet = dwl(LD).

Fig. 8 shows the timing diagram for reactive FPMIPv6 han-dover. Similar to reactive FMIPv6 handover, it is executed whenan MN changes its point of attachment to an nMAG before thefast handover preparation between the pMAG and the nMAGis completed. In other words, reactive FPMIPv6 is performedwhen the MN attaches to the nMAG before the bidirectionaltunnel between the pMAG and the nMAG is established.

Suppose that L(Re-FPMIPv6)HO is the handover latency of reac-

tive FPMIPv6. Then, it is expressed as follows:

L(Re-FPMIPv6)HO = TL2 + TRE-P (27)

where TRE-P is included for representing the times to setupthe IP-layer connection, exchange the required informationbetween the relevant MAGs, and receive the first data packetsent from the nMAG. Note that the data packet is tunneled


from the pMAG to the nMAG and then sent to the MN. Thatis, TRE-P is expressed as follows:

TRE-P =Dπ + dwd(LHI, hA−A) + dwd(LHAck, hA−A)

+ dbuff-packet. (28)

F. Handover Blocking Probability

In order to analyze the handover failure for each mobilitymanagement protocol, the handover blocking probability pre-sented in [25], [29], and [30] is used here. The handover foran MN can fail for several reasons such as unacceptably highhandover latency, signal-to-noise deterioration, and unavailablewireless channel resource. For instance, if the residence timethat the MN stays in the network is less than the handovercompletion time, the handover for the MN is failed due to theloss of the link information or the wireless channel.

Suppose that L(·)HO denotes the handover latency for a specific

mobility management protocol developed in the previous sec-tions. Note that · is used as a protocol indicator. Let E[L

(·)HO] be

the mean value of L(·)HO. Suppose that TR is the residence time

in the network with its probability density function fR(t). Forthe sake of simplicity, L(·)

HO is also assumed to be exponentially

distributed with the cumulative function F(·)T (t). Then, assum-

ing that L(·)HO is the only handover blocking factor, the handover

blocking probability ρb is expressed as follows:

ρb =Pr(L(·)HO > TR

)=

∞∫0

(1− F

(·)T (u)

)fR(u)du

=μcE

[L(·)HO

]

1 + μcE[L(·)HO

] (29)

where μc is the border crossing rate for the MN. Assumingthat the AR’s coverage area is circular, then, μc is calculatedas follows [13], [18], [20]:

μc =2ν

πR(30)

where ν is the average velocity of the MN and R is the radiusof the AR’s coverage area.

G. Packet Loss

While an MN experiences its handover, data packets destinedfor the MN will be lost if any buffer management at networksides does not exist. The amount of packet loss ϕ

(·)p during a

handover is defined as the sum of all lost data packets sent froma CN of the MN. Then, it is expressed as follows:

ϕ(·)p = λsE(S)L

(·)HO (31)

where λs is the average session arrival rate at the MN’s wirelessinterface and E(S) is the average session length in packets. Aspresented in (31), ϕ(·)

p is directly proportionate to L(·)HO. For fast

handover protocols such as FMIPv6 and FPMIPv6, the packet

Fig. 9. Handover latency versus ρf with Dwl = 10 ms.

loss will not occur owing to packet buffering facilities, but onlydelayed packet transportation will occur [13].

V. NUMERICAL ANALYSIS RESULTS AND DISCUSSIONS

In this section, the performance evaluation results of themobility management protocols are presented. For the numer-ical analysis, the following system parameter values are used[25]–[27], [31]: hC−H = 4, hC−G = 6, hH−G = 4, hG−A =4, hA−M = 1, E(S) = 10, τ = 20 ms, n = 3, Lf = 19 B,Dwl = [10, 40] ms, Dwired = 0.5 ms, BWwired = 100 Mbps,TL2 = 45.35 ms, and TDAD = 1000 ms.

A. Handover Latency

Let ρf vary from 0 to 0.7 with a step value of 0.05. Figs. 9 and10 show the handover latency against ρf . A higher value of ρfincreases the probability of the erroneous packet transmissionover the wireless link. Accordingly, the number of mobility sig-naling retransmissions is increased, which results in increasedhandover latency. In other words, as shown in Figs. 9 and 10,the handover latency for each mobility management protocol isrelative to ρf . The value of Dwl also contributes to the handoverlatency. For instance, the handover latency is dramaticallyincreased as the value of ρf is increased with a higher valueof Dwl. Predictive FMIPv6 and FPMIPv6 outperform the othermobility management protocols in terms of handover latencyin this analysis. This is because an MN in those predictivefast handover protocols utilizes the L2 trigger and preparesits handover at the previous (current) access network before itactually moves to the new access network. However, reactivefast handover protocols cannot significantly reduce the han-dover latency because an MN in those protocols must performsome actions at the new access network. Accordingly, fromthese results, it is confirmed that the reactive fast handoverprotocols such as reactive FMIPv6 and FPMIPv6 can be used toprevent packet loss but not to significantly reduce the handoverlatency. Then, PMIPv6 is placed second in this analysis. AnMN in PMIPv6 is locally managed, and mobility signaling is


Fig. 10. Handover latency versus ρf with Dwl = 40 ms.

Fig. 11. Handover blocking probability versus ρf .

exchanged by the LMA and the MAG. It means that mobilitysignaling over the wireless link is not performed so that theeffects of ρf and Dwl are minimized in the performance ofPMIPv6.

B. Handover Blocking Probability

Here, ν and R are set as 20 m/s and 500 m, respectively.Then, Dwl is fixed at 10 ms, while ρf is varied from 0 to 0.7with a step value of 0.05. Fig. 11 shows the handover blockingprobability for each mobility management protocol. Recall thatthe conducted analysis for handover blocking probability onlyconsiders the handover latency as a blocking factor. Similar tothe results shown in Figs. 9 and 10, the handover blocking prob-ability is increased as the value of ρf is increased. The handoverblocking probabilities of predictive FMIPv6 and FPMIPv6are lower than the others as well, but the handover blockingprobability of MIPv6 is higher than the others. PMIPv6 againplaces second in this analysis. Now, ρf and R are set as 0.2

Fig. 12. Handover blocking probability versus ν.

and 500 m, respectively. Then, ν is varied from 0 to 30 m/s.Fig. 12 shows the handover blocking probability against ν. As νis increasing, the MN quickly changes its point of attachments.It means that the MN with a high value of ν is required tocomplete its handover in a shorter time than the MN with alow value of ν. Accordingly, as the value of ν is increased, thehandover blocking probability for each mobility managementprotocol is also increased. In the given analysis environment,only two predictive fast handover protocols such as predictiveFMIPv6 and FPMIPv6 provide good performance in termsof the handover blocking probability that is less than 0.05even if ν is increased until 30 m/s. Note that PMIPv6 alsoshows a considerable performance, i.e., the handover blockingprobability is less than 0.1 when ν reached to 30 m/s. Similarto the previous results, MIPv6 calls forth poor performance interms of the handover blocking probability. This phenomenongets larger as the value of ν is increased. Next, ν and ρf areset as 20 m/s and 0.2, respectively. Then, Dwl is fixed at 10 ms,while R is varied from 400 to 800 m with a step value of 50 m.The high value of R means that the size of the access networkfor the MN is bigger than the low value of R. As R is increased,the residence time, which the MN stays in the access network,is increased so that the MN has more time to complete itshandover while reducing the handover blocking probability. Asshown in Fig. 13, most of the mobility management protocolsare under the influence of R, but R cannot have influence uponthe performance of predictive FMIPv6 and FPMIPv6.

Throughout the results shown in Figs. 11–13, it is confirmedthat the handover latency of predictive FMIPv6 and FPMIPv6is short enough to avoid the handover blocking issues causedby ρf , ν, and R. The reason why those predictive fast handoverprotocols achieve such superior performance compared with theothers is that those protocols allow the MN prepare its handoverat the previous access network before the MN performs theactual handover to the new access network by utilizing the L2information. Regarding the performance of PMIPv6, PMIPv6avoids that mobility signaling, i.e., PBU and PBAck messages,flies on the wireless link so that the value of ρf is not the influ-ence on the performance of PMIPv6. In addition, in PMIPv6,


Fig. 13. Handover blocking probability versus R.

Fig. 14. Packet loss versus ρf with Dwl = 10 ms.

mobility signaling is only exchanged between the MAG andthe LMA over the wired link. Similar to that of MIPv6, thehandover performance of PMIPv6 has been improved as fasthandover techniques were applied, i.e., FPMIPv6.

C. Packet Loss

Without any buffering mechanism, data packets sent from theCN to the MN will be lost while the MN performs its handover.Figs. 14 and 15 show the packet loss during a handover. Here,λs and E(S) are set as one and ten, respectively. Then, ρf isvaried from 0 to 0.7 with different values of Dwl. In Fig. 14,Dwl is set as 10 ms, whereas Dwl is set as 40 ms in Fig. 15.According to the results shown in Figs. 14 and 15, it can beseen that ρf with the higher value of Dwl has more impacton the packet loss. The packet loss during the handover isdirectly proportional to the handover latency as analyzed in theprevious section. For instance, MIPv6 causes a number of lostpackets compared to the others because it requires more timeto complete its handover than the others. Another interesting

Fig. 15. Packet loss versus ρf with Dwl = 40 ms.

observation is that fast handover protocols such as FMIPv6 andFPMIPv6 provide no packet loss during the handover. This isbecause such mobility management protocols adopt the packetbuffering mechanism at ARs/MAGs. For instance, in predictiveFMIPv6, data packets destined for the MN are first forwardedto the nAR, and then, the nAR buffers the data packets untilthe MN arrives at the access network managed by the nAR.As the MN arrives, the nAR forwards the buffered data packetsto the MN. Similarly, data packets sent from the CN to the MNare buffered at the pAR in reactive FMIPv6. Then, when thepAR receives the FBU message indicating that the MN has beenattached to the nAR, the buffered data packets are forwarded tothe nAR. PMIPv6 also yields packet loss even if its handoverlatency is quite low. This is because PMIPv6 does not provideany buffering mechanism to prevent packet loss when the MNperforms its handover.

VI. CONCLUSION

In this paper, the existing IPv6 mobility management proto-cols developed by the IETF have been analyzed and comparedin terms of handover latency, handover blocking probability,and packet loss. From the conducted analysis results, the fol-lowing are confirmed.

1) Utilizing L2 information: In order to improve the han-dover performance, L2 information should be utilized.As shown in Fig. 10, predictive FMIPv6 and FPMIPv6outperform the other mobility management protocols be-cause those protocols allow an MN to prepare its han-dover before the MN performs its actual handover to thenew access network. The reduced handover latency alsoresults in the reduced handover blocking probability asshown in Figs. 11–13.

2) Employing buffering management: In order to preventpacket loss during the handover, any buffering mecha-nism should be employed. As shown in Figs. 14 and15, only fast handover protocols such as FMIPv6 andFPMIPv6 prevent the loss of data packets sent fromthe CN.


3) Wireless link condition: As shown in Figs. 9–11, 14,and 15, the wireless link condition, i.e., FER over thewireless link, largely affects the handover performanceof all mobility management protocols. With this point inview, the network-based mobility management protocolssuch as PMIPv6 and FPMIPv6 have an advantage owingto removed mobility signaling from the MN.

4) DAD latency: As shown in Figs. 10, 14, and 15, MIPv6and HMIPv6 show poor handover performance. Thisphenomenon is caused by the DAD process, which countsfor a large portion of handover latency. Since the DADprocess is performed over a wireless link, in a poorwireless link condition, it badly influences the handoverperformance of MIPv6 and HMIPv6. As a considerablesolution for this, the optimistic DAD [32] is recom-mended that eliminates the DAD completion time.

5) Network topology: As mobility signaling, i.e., BU/BAck,LBU/LBAck, PBU/PBAck, HI/HAck, etc., is sent alongthe network topology, the handover performance isaffected by the network topology configuration. Forinstance, the handover performance of fast handover pro-tocols such as FMIPv6 and FPMIPv6 is largely affectedby the number of hops between the relevant ARs/MAGs.

The conducted analysis results in this paper can be usedto identify each mobility management protocol’s characteris-tics and performance indicators. They could also be used tofacilitate decision making in development for a new mobilitymanagement protocol. For instance, the IETF has recentlyopened the distributed mobility management (DMM) workinggroup aiming at distributing mobile Internet traffic in an optimalway while not relying on centrally deployed mobility anchorssuch as HA, MAP, and LMA. As the DMM approach is inan early stage of standardization, proposals are required to becarefully analyzed and evaluated.

ACKNOWLEDGMENT

This paper is an extension of the first author’s Ph.D. disserta-tion [33]. The companion paper, which was also part of the firstauthor’s Ph.D. dissertation, presents an analytical cost modelfor evaluating the performance of IPv6 mobility managementprotocols [19].

REFERENCES

[1] L. Barolli and F. Xhafa, “JXTA-OVERLAY: A P2P platform for dis-tributed, collaborative and ubiquitous computing,” IEEE Trans. Ind.Electron., vol. 58, no. 6, pp. 2163–2172, Jun. 2011.

[2] A. B. Waluyo, W. Rahayu, D. Taniar, and B. Srinivasan, “A novel structureand access mechanism for mobile broadcast data in digital ecosystems,”IEEE Trans. Ind. Electron., vol. 58, no. 6, pp. 2173–2182, Jun. 2011.

[3] J. Arnedo-Moreno, K. Matsuo, L. Barolli, and F. Xhafa, “Secure commu-nication setup for a P2P based JXTA-Overlay platform,” IEEE Trans. Ind.Electron., vol. 58, no. 6, pp. 2086–2096, Jun. 2011.

[4] L. Han, J. Wang, X. Wang, and C. Wang, “Bypass flow-splitting for-warding in FISH networks,” IEEE Trans. Ind. Electron., vol. 58, no. 6,pp. 2197–2204, Jun. 2011.

[5] Z. An, H. Zhu, X. Li, C. Xu, Y. Xu, and X. Li, “Nonidentical linearpulse-coupled oscillators model with application to time synchronizationin wireless sensor networks,” IEEE Trans. Ind. Electron., vol. 58, no. 6,pp. 2205–2215, Jun. 2011.

[6] B.-S. Choi, J.-W. Lee, J.-J. Lee, and K.-T. Park, “A hierarchical algorithmfor indoor mobile robots localization using RFID sensor fusion,” IEEETrans. Ind. Electron., vol. 58, no. 6, pp. 2226–2235, Jun. 2011.

[7] D. Johnson, C. Perkins, and J. Arkko, “Mobility Support in IPv6,” InternetSoc., Reston, VA, IETF RFC 3775, Jun. 2004.

[8] R. Koodli, “Fast Handovers for Mobile IPv6,” Internet Soc., Reston, VA,IETF RFC 4068, Jul. 2005.

[9] H. Soliman, C. Castelluccia, K. ElMalki, and L. Bellier, “HierarchicalMobile IPv6 Mobility Management (HMIPv6),” Internet Soc., Reston,VA, IETF RFC 4140, Aug. 2005.

[10] R. Koodli, “Mobile IPv6 Fast Handovers,” Internet Soc., Reston, VA,IETF RFC 5568, Jul. 2009.

[11] H. Soliman, C. Castelluccia, K. ElMalki, and L. Bellier, “HierarchicalMobile IPv6 (HMIPv6) Mobility Management,” Internet Soc., Reston,VA, IETF RFC 5380, Oct. 2008.

[12] X. Perez-Costa, M. Torrent-Moreno, and H. Hartenstein, “A performancecomparison of Mobile IPv6, Hierarchical Mobile IPv6, fast handovers forMobile IPv6 and their combination,” ACM SIGMOBILE Mobile Comput.Commun. Rev., vol. 7, no. 4, pp. 5–19, Oct. 2004.

[13] C. Makaya and S. Pierre, “An analytical framework for performanceevaluation of IPv6-based mobility management protocols,” IEEE Trans.Wireless Commun., vol. 7, no. 3, pp. 972–983, Mar. 2008.

[14] S. Gundavelli, K. Leung, V. Devarapalli, K. Chowdhury, and B. Patil,“Proxy Mobile IPv6,” Internet Soc., Reston, VA, IETF RFC 5213,Aug. 2008.

[15] H. Yokota, K. Chowdhury, R. Koodli, B. Patil, and F. Xia, “Fast Han-dovers for Proxy Mobile IPv6,” Internet Soc., Reston, VA, IETF RFC5949, Sep. 2010.

[16] K.-S. Kong, W. Lee, Y.-H. Han, and M.-K. Shin, “Handover latencyanalysis of a network-based localized mobility management protocol,” inProc. IEEE ICC, May 2008, pp. 5838–5843.

[17] J.-H. Lee, Y.-H. Han, S. Gundavelli, and T.-M. Chung, “A comparativeperformance analysis on Hierarchical Mobile IPv6 and Proxy MobileIPv6,” Telecommun. Syst., vol. 41, no. 4, pp. 279–292, Aug. 2009.

[18] J.-H. Lee and T.-M. Chung, “How much do we gain by introducingroute optimization in Proxy Mobile IPv6 networks?” Ann. Telecommun.,vol. 65, no. 5/6, pp. 233–246, Jun. 2010.

[19] J.-H. Lee, T. Ernst, and T.-M. Chung, “Cost analysis of IP mobility man-agement protocols for consumer mobile devices,” IEEE Trans. Consum.Electron., vol. 56, no. 2, pp. 1010–1017, May 2010.

[20] X. Zhang, J. Castellanos, and A. Campbell, “P-MIP: Paging extensionsfor mobile IP,” ACM Mobile Netw. Appl., vol. 7, no. 2, pp. 127–141,Mar. 2002.

[21] J.-H. Lee, S. Pack, I. You, and T.-M. Chung, “Enabling a pagingmechanism in network-based localized mobility management networks,”J. Internet Technol., vol. 10, no. 5, pp. 463–472, Oct. 2009.

[22] H. Fathi, S. S. Chakraborty, and R. Prasad, “Optimization of MobileIPv6-based handovers to support VoIP services in wireless heteroge-neous networks,” IEEE Trans. Veh. Technol., vol. 56, no. 1, pp. 260–270,Jan. 2007.

[23] G. Bao, “Performance evaluation of TCP/RLP protocol stack over CDMAwireless link,” Wireless Netw., vol. 2, no. 3, pp. 229–237, Sep. 2006.

[24] N. Banerjee, K. Basu, and S. K. Das, “Hand-off delay analysis inSIP-based mobility management in wireless networks,” in Proc. IPDPS,Nice, France, Apr. 2003.

[25] S. Yang, H. Zhou, Y. Qin, and H. Zhang, “SHIP: Cross-layer mobilitymanagement scheme based on session initiation protocol and host identityprotocol,” Telecommun. Syst., vol. 42, no. 1/2, pp. 5–15, Oct. 2009.

[26] N. Banerjee, W. Wu, S. K. Das, S. Dawkins, and J. Pathak, “Mobilitysupport in wireless Internet,” IEEE Wireless Commun., vol. 10, no. 5,pp. 54–61, Oct. 2003.

[27] Y.-H. Han, Min, and S.-G. Min, “Performance analysis of HierarchicalMobile IPv6: Does it improve Mobile IPv6 in terms of handover speed?”Wireless Pers. Commun., vol. 48, no. 4, pp. 463–483, Mar. 2009.

[28] S. Thomson, T. Narten, and T. Jinmei, “IPv6 Stateless Address Autocon-figuration,” Internet Soc., Reston, VA, IETF RFC 4862, Sep. 2007.

[29] J. McNair, I. F. Akyildiz, and M. D. Bender, “An inter-system handofftechnique for the IMT-2000 system,” in Proc. INFOCOM, Mar. 2000,pp. 208–216.

[30] J.-H. Lee, M. Kim, B.-S. Koh, and T.-M. Chung, “Adaptive authenticationand registration key management scheme based on AAA architecture,”Intell. Autom. Soft Comput., vol. 16, no. 4, pp. 519–536, 2010.

[31] S. Pack, J. Choi, T. Kwon, and Y. Choi, “Fast-handoff support in IEEE802.11 wireless networks,” IEEE Commun. Surveys Tuts., vol. 9, no. 1,pp. 2–12, 2007.

[32] N. Moore, “Optimistic Duplicate Address Detection,” Internet Soc., Re-ston, VA, IETF RFC 4429, Apr. 2006.

[33] J.-H. Lee, “Enabling network-based mobility management in next-generation all-IP networks: Analysis from the perspective of security andperformance,” Ph.D. dissertation, Sungkyunkwan Univ., Seoul, Korea,Feb. 2010.


Jong-Hyouk Lee (M’07) received the B.S. degreein information system engineering from DaejeonUniversity, Daejeon, Korea, in 2004 and the M.S. andPh.D. degrees in computer engineering under Prof.Tai-Myoung Chung from Sungkyunkwan University,Suwon, Korea, in 2007 and 2010, respectively.

He joined the IMARA team at INRIA,Rocquencourt, France, in 2009, where he workedfor the GeoNet European project, the ITSSv6European project, the MobiSeND French nationalproject, and the SCOREF French national project.

He started his academic profession at the Networks, Security and Multimedia(RSM) Department, TELECOM Bretagne, Cesson-Sévigné, France, in 2012as an Assistant Professor. He is involved in standardization activities atISO TC204 WG16, ETSI TC ITS, and the IETF. He is an Associate Editorof Wiley Security and Communication Networks. His research interestsinclude authentication, privacy, and quality of service in mobile networks;mobility management for vehicular networks; and protocol-operation-basedperformance analysis.

Dr. Lee was the recipient of two Excellent Research Awards from the Depart-ment of Electrical and Computer Engineering, Sungkyunkwan University. He isa member of the Editorial Board of the IEEE TRANSACTIONS ON CONSUMER

ELECTRONICS.

Jean-Marie Bonnin (SM’09) received the Ph.D.degree in computer science from the University ofStrasbourg, Strasbourg, France, in 1998.

He has been with TELECOM Bretagne, Cesson-Sévigné, France, since 2001, where he is currentlythe Head of the Networks, Security and Multime-dia (RSM) Department. His main research interestslie in the convergence between IP networks andmobile telephony networks and particularly in het-erogeneous handover issues. Recently, he has beeninvolved in projects dealing with network mobility

and its application to intelligent transportation systems. He is involved inseveral collaborative research projects at the French and European levels andthrough international academic collaborations (mainly with Asia and NorthAfrica).

Ilsun You received the M.S. and Ph.D. degrees incomputer science from Dankook University, Yongin,Korea, in 1997 and 2002, respectively.

He was with Thin Multimedia Inc., Internet Se-curity Company, Ltd., and Hanjo Engineering Com-pany, Ltd., as a Research Engineer from 1997 to2004. He has been an Assistant Professor with theSchool of Information Science, Korean Bible Uni-versity, Seoul, Korea, since March 2005. He is inthe Editorial Board for the International Journalof Ad Hoc and Ubiquitous Computing, Computing

and Informatics (CAI), the Journal of Wireless Mobile Networks, UbiquitousComputing, and Dependable Applications (JoWUA), the International Journalof Space-Based and Situated Computing, and the Journal of Korean Societyfor Internet Information (KSII). He has served as a Guest Editor of severaljournals such as CAI, MIS, AutoSoft, CAMWA, and WCMC. His main researchinterests include Internet security, authentication, access control, Mobile IPv6,and ubiquitous computing.

Dr. You is a member of IEICE, KIISC, KSII, KIPS, and IEEK. He has servedor is currently serving on the organizing or program committees of internationalconferences and workshops.

Tai-Myoung Chung (SM’00) received the B.S. de-gree in electrical engineering from Yonsei Univer-sity, Seoul, Korea, in 1981, the B.S. degree incomputer science and the M.S. degree in computerengineering from the University of Illinois, Chicago,in 1984 and 1987, respectively, and the Ph.D. degreein computer engineering from Purdue University,West Lafayette, IN, in 1995.

He is currently a Professor with SungkyunkwanUniversity, Suwon, Korea. His research interests areinformation security, information management, and

protocols in next-generation networks.Dr. Chung is currently the Vice-Chair of the Organisation for Economic

Co-operation and Development Working Party on Information Security andPrivacy. He serves as a Presidential Committee member of the Koreane-government and the Chair of the Information Resource Management Com-mittee of the e-government. He is also an expert member of the PresidentialAdvisory Committee on Science and Technology of Korea and is the Chair ofthe Consortium of Computer Emergency Response Teams.

Internet Protocol Version 6

Documents

Transcript of Internet Protocol Version 6