184

JOURNAL OF NETWORKS, VOL. 4, NO. 3, MAY 2009

A New Soft Handover Mechanism using DCHs in 3GPP HSDPA Networks

TaeHoon Lee, SungHoon Seo, UiTaek Lee, and JooSeok Song
Department of Computer Science, Yonsei University, Seoul, Korea Email: {leeth65, hoon, dtmaster, jssong}@emerald.yonsei.ac.kr

AbstractIn 3GPP (Third Generation Partnership Project) Release 5 (R5), HSDPA (High-Speed Downlink Packet Access) has been standardized to enhance the data rate of downlink packet transmission on top of WCDMA (Wideband Code Division Multiple Access) network. Even HSDPA supports up to 10 Mbps data rate through HSDSCHs (High Speed-Downlink Shared Channels), an UE (User Equipment) suffers from significant handover latency since HSDPA standard lacks supporting soft handover. Therefore, in this paper, we propose a new soft handover mechanism for HSDPA to reduce the handover latency by utilizing DCH (Dedicated Channel) instead of HS-DSCH. Simulation results show that the proposed soft handover mechanism reduces the packet IAT (Inter Arrival Time) less than 15 msec while HSDPAs hard handover suffers from 190 msec packet IAT. It means that the proposed mechanism guarantees the minimum packet IAT requirement of the typical VoIP service in 20 msec. Furthermore, the proposed mechanism employees no service disruption for the data connection while an UE performs the soft handover procedure. Index TermsSoft Handover, 3GPP, HSDPA, HS-DSCH, DCH, WCDMA

I. INTRODUCTION WCDMA radio access technology has rapidly evolved to support High-Speed Packet Data Access (HSPDA) for users. Recently, 3GPP R5 introduced HSDPA as a radio access protocol for UMTS (Universal Mobile Telecommunication Standard) to enhance the downlink packet data rate [9][8]. The HSDPA includes a number of new enhanced features such as an adaptive modulation/coding (QPSK/16QAM), a physical layer retransmission (Hybrid-ARQ), a high-speed medium access control layer (MAC-hs) at Node-B, etc [7]. In the HSDPA, downlink packet data is mandatorily transported through HS-DSCH while uplink packet is transported through DCH or HS-DPCCH (HS-Dedicated Physical Control Channel). For the uplink data, UEs can simultaneously transmit data packet via uplink DPCCH with maximum six uplink DPDCHs (Dedicated Physical Data Channels) and a HS-DPCCH. In order to provide compatibility, 3GPP R5 was designed to support the co Correspondence to JooSeok Song, C505, 3rd Engineering Hall, Department of Computer Science, Yonsei University, Shinchon-dong, Seodaemun-gu, Seoul, 120-749, Korea

existence of channels in HSDPA with DCH of the previous release of 3GPP standard (R99). In other words, a Node-B can operate both high-speed channel code in R5 and DCH in R99 at the same time. Then UEs associated with the Node-B can receive packet data via the HS-DSCH for a high-speed or the DCH for a regular speed transmission rate. From the nature of the 3GPP R5, the HSPDA does not support soft handover when UEs move across the multiple Node-Bs so that it may cause a severe throughput degradation and a high handover latency [8]. Typical VoIP applications require that minimum interarrival time (IAT) between every two incoming packets should be less than 20 msec. Since the latency of the hard handover is about 300 500 msec, it cannot guarantee the requirement of the VoIP applications. Therefore, in this paper, we propose a new soft handover mechanism for HSDPA to reduce the handover latency under 20 msec. The proposed mechanism utilizes DCH as a supplementary channel to support soft handover. When an UE performs an inter Node-B handover, from a serving Node-B to a drift Node-B, utilizing the DCH enable the UE to continue its data reception gracefully through the HS-DSCH. The UE establishes a DCH and continues receiving data from the serving Node-B by switching the HSDSCH into the DCH. Then, the UE performs soft handover destined to a drift Node-B by using general DCH handover mechanism, and changes back the current data reception to the HS-DSCH if the drift Node-B supports the HSPDA. We evaluate our proposed soft handover mechanism with ns-2 simulator to compare the performance with existing hard handover mechanism [13][12] in terms of handover latency. In addition, we show that the proposed mechanism minimizes the handover latency to guarantee the requirement of VoIP applications, less than 20 msec.

Figure 1. Hard handover procedure and required channels in 3GPP R5

2009 ACADEMY PUBLISHER

JOURNAL OF NETWORKS, VOL. 4, NO. 3, MAY 2009

185

II. RELATED WORK A. DCH Utilization in R99 and R5 In R99 standard, there exist three different channels for downlink packet data transmission, DCH, FACH (Forward Access Channel) and DSCH (Downlink Shared Channel). As mention in Section I, Node-Bs supporting R5s HSDPA is capable to operate the DCH with highspeed channels in parallel. Furthermore, in R99, the DCH works as a key role for the soft handover since it provides radio bearer to transport signaling for both circuit and packet based services. However, the DCH is not used for the hand handover in R99 any more, since the R5 uses the HS-DSCH as a dedicated channel for hard handover. Fig. 1 shows the hard handover procedure in R5 and the channel utilizations. An UE currently activates a downlink (DL) session through HS-DSCH in a serving Node-B and is going to move to a new drift Node-B in a hard handover area. In the handover area, UEs should perform the hard handover by switching their HS-DSCH in the serving Node-B to the HS-DSCH in the drift NodeB but it causes the severe service degradation of data transmission. Even the R5 provides FCS (Fast Cell Selection) mechanism [8] to avoid the service degradation, the handover latency is too long to guarantee continuous transmission for the delay sensitive applications, e.g., VoIP and streaming. B. Hard Handover Procedure in R5 3GPP R5 standard introduces three handover mechanisms [1][2] as follows: inter Node-B, intra NodeB, and HS-DSCH to DCH handover as shown in Table I. Both inter and intra Node-B handovers are performed from HS-DSCH to HS-DSCH. The inter Node-B handover is a hard handover mechanism thus a large handover latency cannot be avoided whenever the handover is performed. The intra Node-B handover is similar to the softer handover in R9 and GSM but it still brings a little handover latency. While both inter and intra Node-B handovers utilize same types of channel, HSDSCH to DCH handover should switch the channel for data transmission with a different type of channel. Among these R5s handover mechanisms, inter Node-

B (inter RNC) synchronized serving HS-DSCH cell change at hard handover mechanism is introduced in [1][4]. The detail procedure shows in Fig. 2. The SRNC (Serving RNC) decides that there is a need for hard handover combined with serving HS-DSCH cell change. It prepares the RNSAP (Radio Network Subsystem Application Part) message Radio Link Setup Request, which is transmitted to the DRNC (Drift RNC). The DRNC allocates radio resources for the new radio link and requests the drift Node-B to establish a new radio link by transmitting the NBAP (Node-B Application Protocol) message Radio Link Setup Request. The drift Node-B allocates resources and starts physical layer reception on the DPCH on the new radio link and responds with the NBAP message Radio Link Setup Response. The target DRNC responds to the SRNC with the RNSAP message Radio Link Setup Response. The DRNC initiates the setup of Iub DCH and HS-DSCH Data Transport bearers to the drift HS-DSCH Node-B using ALCAP (Access Link Control Application Part) protocol. The SRNC initiates the setup of Iur DCH and HS-DSCH Data Transport bearers. The SRNC transmits the RRC message Physical Channel Reconfiguration to the UE. At the indicated activation time the UE abandons the current active set and initiates establishment of the DPCH in the target cell. When physical layer synchronization is established in the target cell, it starts DPCH reception and transmission and HS-DSCH reception in the target cell. The UE returns the RRC message Physical Channel Reconfiguration Complete to the SRNC. The SRNC then finalizes the procedure by transmitting the RNSAP message Radio Link Deletion Request to the source DRNC. The source DRNC transmits the NBAP message Radio Link Deletion Request to the source Node-B. The source Node-B releases resources for the source radio link and returns the NBAP message Radio Link Deletion Response to the source DRNC. The source DRNC returns the RNSAP message Radio Link Deletion Response to
TABLE I. SUMMARY OF HSDPA HANDOVER TYPES AND THEIR CHARACTERISTICS Intra Node-B Inter Node-B HS-DSCH HS-DSCH to HS-DSCH to to DCH HS-DSCH HS-DSCH Handover Typically by UE, Measurement but possibly also by Node-B Handover By serving RNC Decision Packets Packets not RLC Packet forwarded forwarded. RLC retransmiss Retransmissi from source retransmissions ion used on MAC-hs to used from from target MAC-hs SRNC SRNC No, when RLC acknowledged mode is used, or No, when Packet when duplicate RLC No Losses is used packets acknowled are sent on RLC ged mode unacknowledged mode HS-DPCCH HS-DPCCH Uplink can use softer received by one HS-DPCCH handover cell

Figure 2. Hard handover signaling flow in R5

2009 ACADEMY PUBLISHER

186

JOURNAL OF NETWORKS, VOL. 4, NO. 3, MAY 2009

the SRNC. The DRNC initiates the release of the old Iub DCH and HS-DSCH Data Transport bearers to the drift HS-DSCH Node-B using ALCAP protocol. The SRNC initiates the release of the old Iur DCH and HS-DSCH Data Transport bearers.
Frame length

TABLE II. HS-DSCH IN R5 VS. DSCH IN R99 HS-DSCH 2 msec 16 (with max 15 codes) QPSK, 16QAM DSCH 10, 20, 40, 80 msec 4 ~ 256 QPSK Power control by DCH HARQ

C. HS-DSCH in 3GPP R5 For R5 specifications, released March 2002, the HSDPA was completed, which is the most significant radio related update since the release of the first version of 3GPP WCDMA specifications. HSDPA is based on distributed architecture where the processing is closer to the air interface at the base station (Node-B) for low delay link adaptation. The key technologies used with HSDPA are Node-B based scheduling for the downlink packet data operation, higher order modulation, adaptive modulation and coding, HARQ, and physical layer feedback of the momentary channel condition. The HSDPA operation is carried on the HS-DSCH, which has fundamental differences when compared to DSCH in R99 and R5. The key differences between the HS-DSCH and the DSCH are shown in Table II and Table III. First, the HS-DSCH uses 2 msec frame length while the DSCH uses various frame length in 10, 20, 40 or 80 msec. Second, fixed spreading factor is typically 16 with maximum of 15 codes, while with DSCH the spreading factor may vary between 4 and 256. Third, HS-DSCH also supports 16 Quadrature Amplitude Modulation (16QAM), in addition to Quadrature Phase Shift Keying (QPSK) of DSCH. Forth, Link adaptation, while DSCH is power controlled with the DCH. Finally, physical layer utilizes combining of retransmissions with HARQ. HSDPA extends the WCDMA bit rates up to 10 Mbps. The higher peak bit rates are obtained with higher order modulation, 16QAM, and with adaptive coding and modulation schemes. The theoretical bit rates are shown in Table IV [3]. The maximum bit rate with QPSK modulation is 5.3Mbps and with 16QAM 10.7Mbps. Without any channel coding, up to 14.4Mbps could be achieved. The
TABLE IV. COMPARISON OF FUNDAMENTAL PROPERTIES OF THE DCH AND HS-DSCH Feature Variable spreading factor Fast power control Adaptive modulation and coding Multi-code operation Physical layer retransmissions BTS-based scheduling and link adaptation HS-DSCH No Yes No Yes No No DSCH No No Yes Yes Yes Yes

Spreading factor Modulation Link adaptation PHY retx

TABLE III. MAXIMUM BIT RATE Modulation Coding Rate 1/4 QPSK 2/4 3/4 2/4 16QAM 3/4 10.7 Mbps Max. Bit Rate 1.8 Mbps 3.6 Mbps 5.3 Mbps 7.2 Mbps

terminal capability classes start from 900kbps and 1.8Mbps with QPSK only modulation and 3.6Mbps with 16QAM modulation. The highest capability class supports the maximum theoretical bit rate of 14.4Mbps. III. PROPOSED SCHEME A. Soft Handover Mechanism Utilizing DCHs in R5 We mentioned that there are three handover methods in R5. The intra Node-B HS-DSCH to HS-DSCH handover is already developed in R5 structure so that we did not consider about this method. In addition, HSDSCH to DCH handover means 3G to 2G handover. Therefore, we do not consider this method, too. Our

Figure 3. The proposed soft handover mechanism

2009 ACADEMY PUBLISHER

JOURNAL OF NETWORKS, VOL. 4, NO. 3, MAY 2009

187

Figure 5. Soft handover signaling flow in the proposed scheme

Figure 4. Handover procedure in the proposed handover scheme utilizing DCH

RNC checks next step. As the last step, RNC searches UEs condition. If UE did not allocate DCH, our proposed scheme did not operate, but if UE allocated DCH, RNC is able to decide to make handover for our proposed scheme. B. The Proposed Soft Handover Procedure Fig. 5 shows that SRNC decides to setup a radio link via a new cell controlled by another RNC. SRNC

scheme is devised by replacing inter Node-B HS-DSCH to HS-DSCH handover, because DCH and HS-DSCH always exist together [3]. One dedicated logical channel can be simultaneously mapped onto DCH and HS-DSCH. When UE decides to perform handover, DCH can be used to update the active handover while HS-DSCH provides HSDPA service during the handover. In other words, our goal is to make handover faster using DCH for getting neighbor cells information before handover when RNC decides handover. In addition, we can still use HSDPA during handover. Fig. 3 shows that channels need for soft handover in proposed scheme. In soft handover area, UE changes communication downlink channel from HS-DSCH or HS-SCCH to DSCH, and uplink channel can keep using DCH without change channel. Therefore, we should calculate the channel change delay in UE. A handover delay will be shown in Section IV. Fig. 4 shows that RNC decides handover type considering Node-Bs conditions. UE is connecting on serving Node-B and user makes use of HSDPA service. If UE is moving somewhere, then RNC decides handover for UE. There are three steps for deciding handover in proposed scheme. First, RNC decides to need inter NodeB or intra Node-B handover considering UEs location. If UE located in same Node-B and need not allocation to other Node-B, UE makes handover using already exist mechanism as intra Node-B handover in R5. On the other hand, RNC decides to make inter Node-B handover. Second, RNC checks whether drift Node-B is available HSDPA service or not. If it is not available, RNC decides HS-DSCH to DCH handover. However, if it is available,
2009 ACADEMY PUBLISHER

Figure 6. Simulation topology and wired and radio link configuration

188

JOURNAL OF NETWORKS, VOL. 4, NO. 3, MAY 2009

Figure 7. Impact of TCP throughput for hard handover in the first simulation

Figure 9. Impact of TCP throughput for hard handover in the second simulation

Figure 8. Impact of TCP throughput for soft handover in the first simulation

Figure 10. Impact of TCP throughput for soft handover in the second simulation

requests DRNC for radio resources by sending RNSAP message Radio Link Setup Request. If this is the first radio link via the DRNC for this UE, a new Iur signalling connection is established. This Iur signaling connection will be used for all RNSAP signalling related to this UE. If requested resources are available, DRNC sends NBAP message Radio Link Setup Request to Node-B. Node-B allocates requested resources. Successful outcome is reported in NBAP message Radio Link Setup Response. DRNC sends RNSAP message Radio Link Setup Response to SRNC. SRNC initiates setup of Iur/Iub Data Transport Bearer using ALCAP protocol. This request contains the AAL2 Binding Identity to bind the Iub Data Transport Bearer to DCH. Node-B achieves uplink sync

on the Uu and notifies DRNC with NBAP message Radio Link Restore Indication. In its turn DRNC notifies SRNC with RNSAP message Radio Link Restore Indication. Node-B and SRNC establish synchronism for the Data Transport Bearer(s) by means of exchange of the appropriate DCH Frame Protocol frames Downlink Synchronization and Uplink Synchronization, relative already existing radio link. Then Node-B starts DL transmission. SRNC sends RRC message Active Set Update to UE on DCCH. UE deactivates DL reception via old branch, activates DL reception via new branch and acknowledges with RRC message Active Set Update Complete. SRNC requests DRNC to de-allocate radio resources by sending RNSAP message Radio Link

2009 ACADEMY PUBLISHER

JOURNAL OF NETWORKS, VOL. 4, NO. 3, MAY 2009

189

Figure 11. Packet IAT vs. TCP sequence number for hard handover in the first simulation

Figure 13. Packet IAT vs. TCP sequence number for hard handover in the second simulation

Figure 12. Packet IAT vs. TCP sequence number for soft handover in the first simulation

Figure 14. Packet IAT vs. TCP sequence number for soft handover in the second simulation

Deletion Request. DRNC sends NBAP message Radio Link Deletion Request to Node-B. Node-B reallocates radio resources. Successful outcome is reported in NBAP message Radio Link Deletion Response. DRNC sends RNSAP message Radio Link Deletion Response to SRNC. IV. PERFORMANCE EVALUATION This section provides the evaluation of the proposed soft handover mechanism in terms of TCP throughput and packet IAT comparing with HSDPA hard handover mechanism.

A. Simulation We evaluated the performance of the proposed soft handover compared with the HSDPAs original hard handover mechanism by using UMTS extensions [11] for ns-2 simulator [6]. Since [11] does not provide handover functionalities, we modify the simulation code to support both the hard and the proposed soft handover mechanisms. Fig. 6 shows an entire network topology for our simulation environment. There are two distinguished links in the topology, wired (solid line) and wireless (dashed line) links. For wired links between a CN (Correspondent Node), GGSN (Gateway GPRS Support Node), SGSN (Serving GPRS Support Node), RNC and

2009 ACADEMY PUBLISHER

190

JOURNAL OF NETWORKS, VOL. 4, NO. 3, MAY 2009

Node-B, we configured the link and node characteristics as denoted in the Fig. 6 with droptail queueing model as in [5][10]. On the other hand, the wireless links between Node-B and UE are configured with realistic value for the HSDPA environment in 384 Kbps for downlink and 64 Kbps for uplink channel data rate. During the simulation, the UE moves towards the drift Node-Bs coverage with the moving speed of 60 Km/h and performs handover from a serving Node-B to a drift Node-B, i.e., intra NodeB handover, in the handover area. To simplify, the serving and the drift Node-Bs are connected to the same RNC in the simulation. Every simulation lasts 100 sec trace time and the UE starts to perform handover procedure exactly on 60 sec after the simulation begins. In addition, the CN generate TCP traffic with the segment length in 512 Bytes and TTI (Transmission Time Interval) in 2 msec. B. TCP Throughput The first part of our evaluation is to investigate the impact of TCP throughput when an UE performs handover. We obtain the average TCP throughput by tracing the simulation results in every 500 msec. In the first simulation, Fig. 7 and Fig. 8 are plots of the average TCP throughput as the simulation time flows for the hard handover mechanism and the proposed soft handover mechanism, respectively. As the first result, the TCP throughput of the hard handover mechanism (Fig. 7) is drastically decreased after the handover occurrence at 60 sec of simulation time. From 60 to 61.5 sec, the TCP throughput is decreased less than 51 Kbps since a number of TCP segments are dropped during the hard handover period. On the other hand, the first result of the proposed soft handover mechanism (Fig. 8) shows the slight degradation of the TCP throughput in average 256 Kbps during the handover period. In the second simulation, Fig. 9 and Fig. 10 are plots of the average TCP throughput as same the first simulation. As the second result, the TCP throughput of the hard handover mechanism (Fig. 9) is also decreased after the handover. Therefore, hard handover has penalties for TCP throughput and UE cannot serve good quality voice or data service. On the other hand, the second result of the proposed soft handover mechanism (Fig. 10) shows also the slight degradation of the TCP throughput. This is because the proposed mechanism continues the current data transmission via the HS-DSCH of the serving Node-B by switching channel to the DSCH. As soon as the handover completes, the UE switches back the channel to the original HS-DSCH of the drift Node-B to recover the transmission with HSDPAs fast data rate. C. Packet Inter Arrival Time (IAT) The second part of our evaluation is to compare the hard handover with the soft handover packet IAT when each handover is performed. We collected 143 packets from each simulation around handover occurred time. In the simulation, Figs. 11 and 12 show plots of the packet IAT following the TCP sequence number for the hard
2009 ACADEMY PUBLISHER

handover and the proposed soft handover mechanism. Most of packet IAT is less than 10 msec so that we depict using the semi-logarithmic plot with logarithmic y-axis. In the first result, the maximum packet IAT of the hard handover mechanism (Fig. 11) is 190 msec at 1136-th TCP sequence packet. On the contrary, the first result of the maximum packet IAT of the proposed soft handover mechanism (Fig. 12) is 15 msec at 1162-th TCP sequence packet. In the second result, the maximum packet IAT of the hard handover mechanism (Fig. 13) is 148 msec at 1106th TCP sequence packet. It means that the hard handover mechanism is not suitable for the typical VoIP services because they requires less than 20 msec packet delay for the highest voice quality. On the other hand, the first result of the maximum packet IAT of the proposed soft handover mechanism (Fig. 14) is 17 msec at 1140-th TCP sequence packet. Every packet has IAT in less than 20 msec so that our proposed mechanism is quite suitable to guarantee the requirement of VoIP services. V. CONCLUSION AND FUTURE WORK In this paper, we proposed a new soft handover mechanism for HSDPA based on DCH. Our mechanism has some advantages as follows: First, we can use the existing soft handovers advantages such as shorter handover delay than hard handover mechanism in R99 because we use already existing DCH as a supplementary channel for the soft handover. Second, our proposed mechanism is suitable for delay sensitive applications such as VoIP and multimedia services. On the other hand, our mechanism has also some disadvantages. First, we should endure the existing soft handovers disadvantages such as heavier handover overhead than hard handover. Second, our proposed mechanism is not suitable for delay insensitive service such as web browsing and data downloading services. In addition, we endure about sharing DCH resources. If users use multiple DCHs, the DCH should share the frequency bandwidth resource. It causes bad quality of DCH performance and user suffers from resource limitation. Moreover, the concept of the DCH will be removed in 3GPP LTE (Long Term Evolution) specification. Therefore, proposed soft handover mechanism is useful only 3GPP HSDPA services. As our future work, we will simulate inter RNC soft handover throughput and packet IAT. In addition, we will consider a handover trigger mechanism to assist both delay sensitive and delay insensitive services. Moreover, for increasing reliance we will simulate other situation to make variety environments. ACKNOWLEDGMENT This work was supported by the Korea Science and Engineering Foundation (KOSEF) grant funded by the Korea government (MOST) (No. R01-2006-000-106140).

JOURNAL OF NETWORKS, VOL. 4, NO. 3, MAY 2009

191

REFERENCES
[1] 3GPP Technical Specification 23.009. Digital cellular telecommunications system (Phase 2+); Universal Mobile Telecommunications System (UMTS); Handover procedures, version 6.4.0 Release 6. March 2006. [2] 3GPP Technical Specification 25.308. Universal Mobile Telecommunications System (UMTS); High Speed Downlink Packet Access (HSDPA); Overall description; Stage 2, version 6.4.0 Release 6. March 2007. [3] 3GPP Technical Specification 25.321. Universal Mobile Telecommunications System (UMTS); Medium Access Control (MAC) protocol specification, version 6.14.0 Release 6. October 2007. [4] 3GPP Technical Specification 25.931. Universal Mobile Telecommunications System (UMTS); UTRAN functions, examples on signalling procedures, version 6.4.0 Release 6. June 2006. [5] R. Esmailzadeh. Broadband Wireless Communications Business. Wiley, 2006. [6] EURANE. Enhanced UMTS Radio Access Network Extensions for NS-2. http://eurane.ti-wmc.nl/eurane. [7] F. Frederiksen and T. Kolding. Performance and modeling of WCDMA/HSDPA transmission/H-ARQ schemes. VTC, pages 472476, Sep. 2002. [8] H. Holma and A. Toskala. HSDPA / HSUPA for UMTS High Speed Radio Access for Mobile Communications. John Wiley and Sons, 2006. [9] H. Holma and A. Toskala. WCDMA for UMTS - Radio Access for Third Generation Mobile Communications. John Wiley and Sons, third edition, 2004. [10] A. Mishra. Performance and Architecture of SGSN and GGSN of General Packet Radio Service (GPRS). In IEEE GLOBECOM 2001, 2001. [11] NS-2. The Network Simulator. http://www.isi.edu/ nsnam/ns. [12] K. Pedersen, T. Lootsma, M. Stottrup, F. Frederiksen, T. Kolding, and P. Mogensen. Network performance of mixed traffic on high speed downlink packet access and dedicated channels in WCDMA. VTC, 6:44964500, 2004. [13] K. I. Perdersen, A. Toskala, and P. E. Mogensen. Mobility Management and Capacity Analysis for High Speed Downlink Packet Access in WCDMA. In IEEE VTC, 2004.

TaeHoon Lee received the B.S degree in information and computer engineering from Ajou University, GyeongGi, Korea, in 2007. He is currently working on M.S degree at Yonsei University, Seoul, Korea. He will work at SAMSUNG Electronics as a RESEARCH and DEVELOPMENT ENGINEER from Feb., 2009. His research interests include mobility management in mobile and wireless communications, specifically 3GPP HSDPA, LTE, and WiBro. SungHoon Seo received M.S. degree in computer science from Yonsei University, Seoul, Korea, in 2002. From 2002 to 2004, he had worked at LG Electronics CDMA Handset Laboratory as a RESEARCH ENGINEER. He had been an INTERN RESEARCHER of Microsoft Research Asia, Bejing, China, from 2006 to 2007. He is currently working on Ph.D. degree in computer science at Yonsei University, Seoul, Korea. His research interests include mobility management in mobile and wireless communications, and interworking in heterogeneous networks. Mr. Seo is a student member in IEEE and IEICE. UiTaek Lee received B.S. degree in communication engineering from Myongji University, GyeongGi, Korea, in 2007. He is currently working on M.S. degree at Yonsei University. His research interests include IP multimedia subsystem, p2psip, and mobile communications.

JooSeok Song received the M.S. degree in electrical engineering from Seoul National University, Seoul, Korea, in 1976, and the M.S. degree in electrical engineering from KAIST, Korea, in 1979. In 1988, he received the Ph.D. degree in computer science from University of California at Berkeley. He had been an ASSISTANT PROFESSOR of Naval Postgraduate School, California, USA, from 1988 to 1989. He is currently a PROFESSOR of computer science at Yonsei University, Seoul, Korea. His research interests include cryptography, network security, and data communications.

2009 ACADEMY PUBLISHER