efficient architecture and handoff strategy used for … personal communications (2007)...

14
Wireless Personal Communications (2007) 43:201–214 DOI 10.1007/s11277-006-9218-3 c Springer 2006 Efficient Architecture and Handoff Strategy used for VoIP Sessions in SIP Based Wireless Networks TIN-YU WU 1 , RONG-WEI JHANG 1 and HAN-CHIEH CHAO 2 1 Department of Electrical Engineering, National Dong Hwa University, Hualien, Taiwan, ROC E-mails: [email protected], [email protected] 2 Department of Electronic Engineering, National Ilan University, Ilan, Taiwan, ROC, E-mail:[email protected] Abstract. In the near future, the Internet is likely to become an All-IP network that provides various multimedia services over wireless networks. Although the earliest VoIP applications did not consider the end-node mobil- ity, researchers have attempted to support mobility in current VoIP protocols, such as Session Initial Protocol (SIP)-based mobility. The SIP-based mobility is considered because it can readily support mobility. However, calling disruptions may occur in traditional SIP mid-call terminal mobility because handoff procedure may be required, depending on the implementation and the real network deployment considerations. In any case, issues in the combined SIP/RSVP for guaranteeing QoS of VoIP service under mobile environment are also considered to be crucial. Therefore, this study describes the solutions by devising novel hierarchy network architecture. Also, the mechanisms including help with neighboring users in adjacent cells and the third party call control to over- come those issues are included. The simulation results indicate that the proposed technique is practical and better executive than conventional schemes. Keywords: VoIP, IP mobility, SIP, RSVP, QoS 1. Introduction Recently, the growth of Internet has led to an All-IP network to provide various multimedia services over wireless nodes. Researchers have attempted to support mobility in current VoIP protocols. Currently, two basic approaches are known for supporting mobility in VoIP services. The first approach attempts to support mobility in the network layer by using Mobile IP and its related proposals, while the other framework relies on the application layer by augmenting existing protocols, such as Session Initial Protocol (SIP). Although mobility for VoIP services can be supported via Mobile IP, there are some potential shortcomings, such as routing ineffi- ciency, overhead problems, handoff latency, and stability problems as pointed out in [1–3]. Additionally, other studies [4, 5] have had demonstrated differences in performances between SIP-based mobility and Mobile IP. Most of them conclude that the Mobile IP approach can support non-real-time sessions while SIP-based mobility better supports real-time sessions, such as VoIP. Therefore, SIP based mobility seems to be the most suitable method to handle VoIP sessions for mobility support. This study presents solutions to lower handoff delay time; thereby, reducing resulted packet loss, RSVP path re-establishment time and resources usage in SIP-based mid-call terminal mobility. Additionally, an RSVP incorporated environment is proposed by devising hierarchy network architecture and methods including helping neighboring users and the third party call control. The aim of the proposed scheme solicits some users in neighboring cells for assisting

Upload: haphuc

Post on 20-Apr-2018

215 views

Category:

Documents


1 download

TRANSCRIPT

Wireless Personal Communications (2007) 43:201–214DOI 10.1007/s11277-006-9218-3 c© Springer 2006

Efficient Architecture and Handoff Strategy used for VoIP Sessionsin SIP Based Wireless Networks

TIN-YU WU1, RONG-WEI JHANG1 and HAN-CHIEH CHAO2

1Department of Electrical Engineering, National Dong Hwa University, Hualien, Taiwan, ROCE-mails: [email protected], [email protected] Department of Electronic Engineering, National Ilan University, Ilan, Taiwan, ROC,E-mail:[email protected]

Abstract. In the near future, the Internet is likely to become an All-IP network that provides various multimediaservices over wireless networks. Although the earliest VoIP applications did not consider the end-node mobil-ity, researchers have attempted to support mobility in current VoIP protocols, such as Session Initial Protocol(SIP)-based mobility. The SIP-based mobility is considered because it can readily support mobility. However,calling disruptions may occur in traditional SIP mid-call terminal mobility because handoff procedure may berequired, depending on the implementation and the real network deployment considerations. In any case, issuesin the combined SIP/RSVP for guaranteeing QoS of VoIP service under mobile environment are also consideredto be crucial. Therefore, this study describes the solutions by devising novel hierarchy network architecture. Also,the mechanisms including help with neighboring users in adjacent cells and the third party call control to over-come those issues are included. The simulation results indicate that the proposed technique is practical and betterexecutive than conventional schemes.

Keywords: VoIP, IP mobility, SIP, RSVP, QoS

1. Introduction

Recently, the growth of Internet has led to an All-IP network to provide various multimediaservices over wireless nodes. Researchers have attempted to support mobility in current VoIPprotocols. Currently, two basic approaches are known for supporting mobility in VoIP services.The first approach attempts to support mobility in the network layer by using Mobile IP andits related proposals, while the other framework relies on the application layer by augmentingexisting protocols, such as Session Initial Protocol (SIP). Although mobility for VoIP servicescan be supported via Mobile IP, there are some potential shortcomings, such as routing ineffi-ciency, overhead problems, handoff latency, and stability problems as pointed out in [1–3].Additionally, other studies [4, 5] have had demonstrated differences in performances betweenSIP-based mobility and Mobile IP. Most of them conclude that the Mobile IP approach cansupport non-real-time sessions while SIP-based mobility better supports real-time sessions,such as VoIP. Therefore, SIP based mobility seems to be the most suitable method to handleVoIP sessions for mobility support.

This study presents solutions to lower handoff delay time; thereby, reducing resulted packetloss, RSVP path re-establishment time and resources usage in SIP-based mid-call terminalmobility. Additionally, an RSVP incorporated environment is proposed by devising hierarchynetwork architecture and methods including helping neighboring users and the third party callcontrol. The aim of the proposed scheme solicits some users in neighboring cells for assisting

202 T.-Y. Wu et al.

the mobile node (MN). The proposed scheme also aims to handle and achieve SIP mid-callterminal mobility procedures involving location update, and rapid RSVP path re-establishmentby using third party call control mechanism during handoff.

Section 2 introduces related works. Section 3 describes the proposed approach and thesimulation model. Section 4 presents the simulation results. The conclusion and the futurework are given in Section 5.

2. Related Works

Herein, pertinent literature is surveyed. In [6], describes the structure of the SIP multicastmobility that is independent of the IP layer and extends SIP functions to support SIP multicastmobility. The proposed method avoids problems exiting in the IP multicast mobility by shift-ing the concept of multicasting and mobility from IP layer to session and application layers.In [3], two composite mobility management architectures based on SIP and Mobile IP areproposed to support multimedia services in a seamless fashion. The first approach uses SIPwith IP encapsulation measures on correspondent node (CN) to support mobility for all trafficfrom or toward the MN. However, the second approach segregates traffic and adopts SIP withthe NAT mechanism to support mobility for real-time traffic over UDP with Mobile IP sup-porting mobility for non-real-time traffic mainly TCP-based applications [3]. Also comparesmobility schemes with a table, which shows the advantages of both the proposed methodsover Mobile IP including the support of personal and session mobility, the lack of a singlepoint of failure, and mobility awareness above the IP layer. Therefore, SIP-based mobilityappears to be practical and possesses many useful features of supporting multimedia servicesin all-IP networks. Yet, whether to use SIP supporting mobility for all services, or to use SIPand Mobile IP for supporting real-time and non-real-time applications, respectively, is a ques-tion that should be further considered. Fortunately, [4, 5] perform simulations or laboratoryexperiments to solve this question. It shows the simulation and experimental results that SIPprovides better throughput and performance than mobile IP owing to latency improvementand increased packet size utilization in real-time (RTP/UDP) traffic.

The study in [7] applies SIP-based mobility mainly to dominate mobility management forVoIP sessions as Mobile IP has some flaws in real-time application. However, calling disrup-tions may still occur in the SIP mid-call mobility. Thus, the proposed mechanism uses MobileIP only to tunnel data packets initiated before the mobile host moves to the new network,while the sessions initiated after moving with the mobile host’s new IP address and takingthe direct path to/from the MN. However, Mobile IP must be integrated with SIP; therefore,the home agent (HA) or SIP proxy should be configured to store both “user address binding”(UserURI, and ContactURI) and “IP address binding” (Home Address, and care-of addressinformation). This configuration may complicate the implementation and design the homeagent or SIP proxy. Moreover, end-users also need to be aware of (what) and support both SIPand Mobile IP operations. However, our proposed method uses only a SIP related mechanismcalled third party call control. With the help from neighboring users, this method can handlethe problems mentioned in previous section, but there is no non-SIP-based mobility approachthat can be added to the system.

In short, previous works clearly show that SIP-based mobility is an appropriate mobilityapproach for supporting real-time multimedia services such as the VoIP service. However, call-ing disruptions may occur during standard SIP mid-call terminal mobility procedures. Also,

Efficient Architecture and Handoff Strategy 203

Table 1. Comparison of mobility management schemes

Mobility approach

Third party

call control

with neigh-

boring

MIP with Multicast users help

route optimi- Hybrid mobility in (proposed

Ability/feature MIP zation SIP-based SIP/MIP SIP layer scheme)

Optimized routing No Yes Yes Yes Yes Yes

Personal and sessionmobility

No No Yes Yes Yes Yes

No IP stack modifica-tions

Yes No Yes No Yes Yes

Seamless handoff sup-port

Possibly Possibly Possibly Possibly Possibly Yes

Suitable for which typeof service

Non-real-timeApp.(TCP-based)

Non-real-time App.(TCP-based)

Real-time App.(UDP-based)

Real-time App.(UDP-based)

Real-time App.(UDP-based)

Real-timeApp.(UDP-based)

Need additional mobil-ity approach other thanitself based added to it

No No No Yes Yes No

Concerning QoS guar-antee also

Possibly (notspecifiedspecifically)

Possibly (notspecifiedspecifically)

Possibly (notspecifiedspecifically)

Possibly (notspecifiedspecifically)

Possibly (notspecifiedspecifically)

Yes

deploying the combined SIP and RSVP architecture can ensure QoS for VoIP applications, butwhen employed in wireless/mobile networks, it may also incur issues, such as RSVP path re-establishment. The next subsection discusses both these issues. Table 1 compares the mobilitymanagement methods discussed in this section.

2.1. P r o b l e m s i n S I P M i d - c a l l T e rminal Mobility

In conventional SIP mid-call terminal mobility [1, 7], as seen in Figure 1, the MN first detectsits movement to the new cell, possibly a new wireless IP subnet, resulting in L3 handoff afterentering the overlapped area. Then, the MN requests a new contact IP address from the DHCPserver, depending on implementation and real network deployment considerations. However,this procedure can contribute significantly to overall handoff delay. Then, the MN can re-invite the CN (Macro-mobility) or re-register to the outbound proxy server (Micro-mobility)[1, 8, 9] to update its location. The new location thus becomes reachable by other hosts afterhanding off to the new cell. Nevertheless, if the new SIP session is not created completelywhile the MN is in the overlapped area [7], calling disruptions may occur. Figure 2 shows thisphenomenon.

204 T.-Y. Wu et al.

Figure 1. Operation of SIP mid-call mobility for micro-mobility.

Figure 2. Timing chart in SIP mid-call terminal mobility.

2.2. I s s ue s i n t he C o mbined SIP and RSVP Architecture

SIP is usually adopted to signal multimedia sessions, like VoIP session. Such real-time multi-media communications are more likely to incorporate QoS guarantee mechanisms. Therefore,engineers believe that some QoS provision mechanisms, such as RSVP, should be incorpo-rated into SIP to collaborate with other communications, as stated in [5]. However, previous

Efficient Architecture and Handoff Strategy 205

literatures do not consider many details that the combined architecture employed in the mobileenvironment scenario and especially in the handoff case. The well-designed RSVP is origi-nally used on fixed-wire networks for resource reservation to guarantee the required QoS onthe path between the sender and recipient. However, RSVP cannot handle mobile hosts chang-ing their IP addresses when performing L3 handoff during a connection lifetime. Therefore,it must initiate a new round of signaling when used in the wireless/mobile environment. Inshort, RSVP path re-establishment and handoff procedures are important issues in SIP-basedmobility environment after SIP and RSVP are integrated in wireless networks.

3. Proposed Approach and Simulation Model

3.1. C o nc e p t i o n o f t he P r oposed Mechanism

First, the proposed scheme to resolve the issues in the combined SIP and RSVP architecture isintroduced as specified in [5], as applied in wireless/mobile networks as discussed above. Thisproposal uses the hierarchy network topology as shown in Figure 3 with a gateway (GW) act-ing as a root that can (enhanced TRY adapted) act as both the SIP Proxy and RSVP router. Theproxy is a mobility-aware entity, which handles micro-mobility within its managed domainas assumed above and in [1, 9] (the MN advertising dose not need own address as the mediaaddress, but rather that of the proxy. Alternatively the proxy can rewrite MN’s contact addressas its own IP address)

It can also be a RSVP-capable router that restricts RSVP path re-establishment and existsonly between the proxy and the MN’s new contact address. This capability is achieved bytranslating the MN’s new contact IP address in the Path and Resv messages into the proxy’sIP address. Or, the reverse procedure can be performed depending on whether messages gooutside or inside the proxy-managed domain. This operation hides the handoff and reducesthe resulted new RSVP signals that are exchanged outside the proxy-managed domain duringhandoff. Also, the unchangeable identity, SIP URI, can be used to identify the user (MN) that is

Figure 3. Architecture of the overall proposal.

206 T.-Y. Wu et al.

Figure 4. Operation of new RSVP signaling in our proposal.

involved in resource path re-establishment. Since, the proxy is a mobility aware entity trackingvalid bindings between the user identifier, SIP URI and its new IP address, it could also be anRSVP-enabled router when utilizing SIP with RSVP in wireless/mobile networks. Typically,these involve in RSVP path re-establishment because the end-user’s IP address changes whenthe MN performs SIP-based mobility mechanism. Thus, it places the new RSVP signal reduc-tion outside the proxy-managed domain. Figure 4 depicts this operation. It clearly reveals thatthe reservation states outside the proxy-managed domain are configured for the stable proxy’sIP addresses, and messages that are sent to outside of the proxy-managed domain serves merelyas state refresh messages. Only the reservations inside the proxy-managed domain are MN’snew dependent IP contact address and are used to re-establish connection. Clearly, introduc-tion of this designation and deployment will have three main advantages that can be brieflydescribed as reducing, duplicate, and resource reservation, that is for the same flow during aperiod of non-negligible time, lowering possibility of dropping the reservation altogether thatis caused by the limitations of resources in the core or other transit networks, minimizing longdelay for path reservation re-establishment.

Second, by designing two schemes helping neighboring users in neighboring cells and thethird party call control, the proposal further aims to reduce the time of handoff delay/disrup-tion and the resulted packet loss in SIP-based mobile networks. The proposed schemes solicitneighboring users for assisting the MN in handling SIP mid-call terminal mobility procedures,including location update and RSVP path re-establishment by using the third-party call controlmechanism during handoff.

Figure 3 indicates that when the MN moves toward a new cell and predicts or finds that itleads to L3 handoff, it will then exchange control messages with neighboring users to selectthe most suitable user to help it commence SIP re-registration and then trigger RSVP pathre-establishment to the Proxy/Enhanced RSVP router. Additionally, during the neighboring

Efficient Architecture and Handoff Strategy 207

Figure 5. Conception and exchanged messages in our proposal.

user selection, the MN and neighboring users operate in ad hoc mode and only exchange con-trol massages to save the originally inadequate resources used for the current VoIP sessionsin wireless networks. Its conception and operation are partly similar to [10], which describesa framework for applying SIP in an ad hoc network to establish multi-party voice sessions.However, in our proposal, it only requires the exchange of control messages to signal the hand-off procedure, so the control messages sending from the MN to neighboring users will mainlycontain the information, such as, the MN’s SIP URI and the MAC address, that is sufficient forthe purposes of neighboring users and avoids wasting time in sending unnecessary informa-tion. Furthermore, when choosing the most appropriate neighboring user, only the candidateusers that meet certain criteria can respond to the MN’s requests and thus can be chosen.After the MN has chosen one neighboring user, the selected user immediately commences thethird-party call control procedure. As described in [1, 8, 11], the third-party call control is afundamental telephony function used by many services, such as call set up by a controller andcall transfers. The third party call control is suitably applied here to re-establish or resumesessions during handoff. It was aided by an additional controller, and the selected neighboringuser is just as the mentioned previously. Therefore, to accomplish the SIP mid-call terminalmobility, the selected user first invites the MN to obtain SIP-related information, such as SIPURI and media parameters described in the SDP reply message sending from the MN. Then, itsends an Ack message back to the MN with hold SDP back to indicate that no media exchangeis possible. Meanwhile, the selected neighboring user which has SIP-related information aboutthe MN needs to re-register with the proxy, but it does not yet have a new IP address requiredby the MN as the contacted address in the new subnet. Therefore, the selected user interactswith the DHCP server to acquire a new IP address for the MN. Then, it re-registers with theproxy for the MN. Next, as described above, the proxy enhanced as an RSVP router, initiatesbi-directional RSVP path re-establishment between the MN’s new contacted address and theproxy, (helped OR aided) by the neighboring user. Then, the updated message indicating thecompletion of the resource re-establishment, and the OK message indicating the proxy agrees

208 T.-Y. Wu et al.

re-registration, are sent by the proxy. Thus, the selected user receives information from theproxy responding to MN’s re-registration solicitation.

Therefore, the selected user re-invites the MN with information that is just received fromthe proxy agreeing to the MN’s re-registration, but it may modify or agree only some parts ofthe SDP that MN wants to negotiate. When it receives the information, the MN could alterits’ desired media types and parameters to match those of the proxy. The MN then sends thisinformation in a final SDP “OK” message back to the selected user. Finally, the selected usersends ACK messages to both the MN and proxy to complete the third party call control trans-action. Clearly, at this moment, the selected user helps to perform procedures required in SIPmid-call terminal mobility by executing the third party call control mechanism to exchange therequired information and messages between the MN and the proxy. Therefore, after handingoff the new subnet, the MN should have sufficient reserved resources to quickly accomplishSIP registration procedures and recover or maintain current VoIP sessions with minimum dis-ruption when deploying VoIP sessions in an SIP based wireless/mobility environment withRSVP. Figure 5 shows the concepts and messages that are exchanged among several entitiesin this proposal.

4. Simulation Results

This section describes the simulations based on our proposal and discusses their results. First,a simulation is set up to measure the dropped packets resulting from the delayed time whenresuming ongoing sessions during the handoff periods. In this simulation, the CN unceas-ingly sends UDP packets every 1 m/sec to the MN that moves through two cells at velocity of15 m/sec. Figure 6 shows that the proposed mechanism can recover the ongoing sessions morequickly, and therefore, it maintains communications with minimum disruption when handoffoccurs.

Figure 6. Received packets sequence number from 3 to 4.5 sec.

Efficient Architecture and Handoff Strategy 209

Figure 7. The amount of packet loss.

Figure 7 also presents the total packet loss measured by counting the correspondingsequenced numbers at each time point. This result also demonstrates that the proposed mech-anism outperforms conventional mechanisms because packet loss is lower in the proposedmechanism.

Next, reservation occupancies about the proposed mechanism concerning on RSVP issuesare discussed. The simulation topology is shown in Figure 8. Here, every RSVP request isconsidered as a request for a telephone call. Since admission control is also assumed to be inplace, circuit-switching terminology, such as call blocking probability, is used. A call requestis blocked when there is no free channel to reserve. Also, this study focuses on the link betweenthe proxy/enhanced RSVP router, the access network edge component and the core router inthe upstream ISP because this link is usually the bottleneck for the access networks, and it isalso an expensive resource. In the simulation, the access routers, the proxy and the core routerare assumed to support 512, 1024, 256 channels, respectively.

Figure 9 depicts the reservation occupancies at the link between the proxy and the corerouter against the new traffic arrival rate as a fraction of the total bandwidth of the link. Theanalytical results show that the proposed mechanism does not create any stale reservationson that link. However, the conventional mechanism retains about 23% of the link bandwidthunnecessarily occupied because of the stale sessions it maintains. Furthermore, the stale-activeratio under the conventional mechanism does not appear to depend on the new traffic arrivalrate. Under the conventional mechanism, both the stale and active reservations increase untiltheir sum reaches the total bandwidth of the link, while the proposed mechanism can use theentire bandwidth to support the active connections.

Further comparisons between those four UDP-based real time protocols as listed in Table 1(SIP-based, Hybrid SIP/MIP, Multicast Mobility in SIP Layer, and Proposed mechanism) andthe proposed one has been conducted. Figure 10 shows what handoff frequency has affectedthe packet lost. Since the handoff latency for every mechanism is greater than 1000 ms, the

210 T.-Y. Wu et al.

Figure 8. Simulation topology about RSVP issues.

Figure 9. Reservation percentage in the link between edge and core router.

packet lost ratio is large in general for handoff interval below 10 sec. Without the RSVP capa-bility, the Multicast Mobility in SIP mechanism still suffers the higher packet lost ratio thanothers even the interval is more than 20 sec.

Figure 11 shows the time needed to establish a RSVP path during handoff. The path issupposed to be based on a five levels of tree structure. At this time, Multicast Mobility in SIPmechanism shows the less time than others for in lack of RSVP capability. For the proposedone, it will not re-establish the path between each level while only the level right above theAR due to its hierarchy nature. It is shown that the latency is reduced in general.

Efficient Architecture and Handoff Strategy 211

0

2

4

6

8

10

12

14

16

5 10 15 20 25 30

Handoff Interval (sec)

Pack

et L

ost R

atio

(%

)

Figure 10. Packet lost ratio versus Handoff interval.

Figure 11. Latency for establish a RSVP path.

5. Conclusion and Future Work

This paper considers issues in deploying VoIP sessions in SIP-based wireless and mobilenetworks, including the inefficient re-establishment of resources in the normal SIP, RSVPcombined architecture, and calling disruptions in the SIP mid-call terminal mobility. To han-dle these problems, this investigation proposes solutions by devising the hierarchical networkarchitecture, and mechanisms including the help from neighboring users and third-party callcontrol. From conception and operation that are discussed and elaborated above, our proposalhas been shown to solicit some neighboring users for assisting the MN in quickly handlingand accomplishing SIP mid-call terminal mobility procedures during the handoff. Addition-ally, the results of the simulation reveal that the proposed hierarchical network architectureand mechanisms outperforms the conventional SIP mid-call terminal mobility approach, andalso combines with RSVP, for example, in terms of handoff delay time, packet loss, and thereservation re-establishment delay. However, the important considerations, such as securityissues for the MN, neighboring users and the proxy need to be further considered in the futureresearches.

212 T.-Y. Wu et al.

Acknowledgements

This work is a partial result of project no NSC 94-2219-E-259-001 and NSC 94-2219-E-259-002 conducted by National Dong Hwa University under the sponsorship of the NationalScience Conucil, Taiwan, ROC.

References

1. H. Schulzrinne and E. Wedlund, “Application-Layer Mobility Using SIP”, ACM Mobile Computing and Com-munication Review, Vol. 1, No. 2, pp. 47–57, 2001.

2. A. Misra, S. Das, and P. Agrawal, “Application-Centric Analysis of IP-based Mobility Management Tech-niques”, Journal of Wireless Communications and Mobile Computing, Vol. 1, No. 3, 2001. Wiley Interscience,New York.

3. K.D. Wong, A. Dutta, J. Burns, R. Jain, K. Young, and H. Schulzrinne, “A Multilayered Mobility ManagementScheme for Auto-Configured Wireless IP Networks”, IEEE Wireless Communications Magazine, 2003.

4. K.D. Wong, A. Dutta, K. Young, and H. Schulzrinne. On handling. in C. Politis, K. Chew, and R. Tafazolli(eds.), “Multilayer Mobility Management for All-IP Networks: Pure SIP vs. Hybrid SIP/Mobile IP”, IEEEVTC, 2003.

5. G. Camarillo, W. Marshall, and J. Rosenberg, “Integration of Resource Management and SIP”, RFC3312,IETF, 2002.

6. X. Yang and A. Agarwal, “Multicst Mobility in SIP Layer”, in Proceeding of Vehicular Technology Conference,2004. VTC 2004-Spring. 2004 IEEE 59th, Vol. 5, pp. 2667–2671, 17–19 May 2004.

7. J.-W. Jung, R. Mudumbai, D. Montgomery, and H.-K. Kahng, “Performance Evaluation of Two LayeredMobility Management using Mobile IP and Session Initiation Protocol”, IEEE GLOBECOM, 2003.

8. H. Schulzrinne and J. Rosenberg, “The Session Initiation Protocol: Internet-Centric Signaling”, IEEE Com-munication Magazine, 2000.

9. T.T. Kwon, M. Gerla and S. Das, “Mobility Management for VoIP Service: Mobile IP vs. SIP”, IEEE WirelessCommunications, 2002.

10. H. Khlifi, A. Agarwal and J.-C. Gregoire, “A Framework to Use SIP in Ad-hoc Networks”, IEEE CCECE,2003.

11. H. Sinnreich and A. B. Johnston, Internet Communications Using SIP, New York: Wiley, 2001.

Han-Chieh Chao is a joint appointed Full Professor of the Department of Electronic Engi-neering and Institute of Computer Science and Information Engineering. He also serves as theDean of the College of Electrical Engineering and Computer Science and Director of Com-puter and IT Center for National Ilan University, I-Lan, Taiwan, ROC. His research interests

Efficient Architecture and Handoff Strategy 213

include High Speed Networks, Wireless Networks, IPv6 based Networks, Digital CreativeArts, and Digital Divide. He received his MS and Ph.D. degrees in Electrical Engineeringfrom Purdue University in 1989 and 1993, respectively. He has authored or co-authored fourbooks and has published about 140 refereed professional research papers. He has completed40 MSEE thesis students. Dr. Chao has received many research awards, including PurdueUniversity SRC awards, and NSC research awards (National Science Council of Taiwan). Healso received many funded research grants from NSC, Ministry of Education (MOE), RDEC,Industrial Technology of Research Institute, Institute of Information Industry and FarEasToneTelecommunications Lab. Dr. Chao has been invited frequently to give talks at national, andinternational conferences and research organizations. Dr. Chao is also serving as an IPv6Steering Committee member and co-chair of R&D division of the NICI (National Informationand Communication Initiative, a ministry level government agency which aims to integratedomestic IT and Telecom projects of Taiwan), Co-chair of the Technical Area for IPv6 ForumTaiwan, the executive editor of the Journal of Internet Technology and the Editor-in-Chief forInternational Journal of Internet Protocol Technology and International Journal of Ad Hoc andUbiquitous Computing. Dr. Chao is an IEEE senior member.

Tin-Yu Wu is currently serving as the Network Division in the University Computer and ITCenter at National Dong Hwa University (NDHU), Hualien, Taiwan, ROC. He received hisMS degree in Electrical Engineering from NDHU in 2000. His research interests focus onthe next generation Internet protocol, mobile computing, and wireless networks. He is now aPh.D candidate in Department of Electrical Engineering, NDHU.

214 T.-Y. Wu et al.

Rong-Wei Jhang received his MS degree in Electrical Engineering from NDHU in August2004. He is currently a software engineer serving in Abocom Systems, Inc., which providessolutions of wireless and broadband network products. His research interests include mobilecomputing and wireless communications.