Design Architectures for 3G and IEEE 802.

11 WLAN Integration
Abstract
Wireless LAN access networks show a strong potential in providing a broadband complement to Third Generation (3G) cellular systems. 3G networks provide a wider service area, and ubiquitous connectivity with low-speed data rates as compared to Wireless Local Area Networks (WLANs). WLAN networks offer higher data rate and the easy compatibility to wired Internet, but cover smaller areas. Integrating 3G and WLAN networks can offer subscribers high-speed wireless data services as well as ubiquitous connectivity. The key issue involved in achieving these objectives is the development of integration architectures of WLAN and 3G technologies. The choice of the integration point depends on a number of factors including handoff latency, mobility support, cost-performance benefit, security, authentication, accounting and billing mechanisms. We review 3G-WLAN integration architectures and investigate two such architectures in the case when the Universal Mobile Telecommunications Network (UMTS) network is connected to a WLAN network at different integration points, namely the Serving GPRS Support Node (SGSN) and the Gateway GPRS Support Node (GGSN). The evaluation of these integration architectures were conducted through experimental simulation tests using OPNET.

1. Introduction
Mobile communications and wireless networks are developing at a rapid pace. Advanced techniques are emerging in both these disciplines. There exists a strong need for integrating WLANs with 3G networks to develop hybrid mobile data networks capable of ubiquitous data services and very high data rates in strategic locations called hotspots. 3G wireless systems such as Universal Mobile Telecommunication Systems (UMTS) can provide mobility over a large coverage area, but with relatively low speeds of about 144 Kbits/sec. On the other hand, WLANs provide high speed data services (up to 11 Mbits/sec with 802.11b) over a geographically smaller area. WLANs are generally used to supplement the available bandwidth and capacity of a 3G network in hotspot area such as railways and airports with high trafficdensities, without sacrificing the capacity provided to cellular users. The rest of this paper is organized as follows. Section 2 provides a brief background on 3G and WLAN networks. Section 3 describes the related research and contributions of this work. Section 4 presents a comparison of the various internetworking architectures. In section 5 we compare two integration architectures connecting UMTS and 802.11 networks. Section 6 presents our simulation results. Finally in section 7 we present some concluding remarks and future works.

2. Background
2.1 WLAN Standards 802.11b [4] WLAN has been widely deployed in offices, homes and public hotspots such as airports and hotels given its low cost, reasonable bandwidth (11Mbits/s), and ease of deployment. However, a serious disadvantage of 802.11 is the small coverage area (up to 100 meters) [7]. Other 802.11 standards include 802.11a and 802.11g which allow bit rates of up to 54 Mbits/sec. 2.2 3G Networks ITU defines 3G as any device that can transmit or receive data at 144Kbps or better [6]. In practice, 3G devices can transfer data at up to 384 Kbps. As a comparison, Global System for Mobile Communications (GSM) data rates are up to 14.4 Kbps and General Packet Radio Service (GPRS) is around 53.6 Kbps used in 2G and 2.5G respectively. Two main proposed systems for 3G recognized by the International Telecommunication Union (ITU) are Code Division Multiple Access (CDMA 2000) and Universal Mobile Telecommunication System (UMTS) [19]. CDMA2000 1x is an evolution of cdmaOne [6], and supports packet data service up to 144 Kbits/sec. CDMA 2000 1xEV-DO introduces a new air interface known as High Rate Packet Data (HRPD) and provides up to 2.4 Mbits/sec on the downlink (from base station to terminal), but only 153 Kbits/sec on the uplink. CDMA2000 1xEV-DV is expected to introduce new radio techniques and an all-IP architecture for radio access and core networks. It promises data rates up to 3 Mbits/sec. UMTS is composed of two different but related modes: Wideband CDMA also called Frequency

Division Duplex (FDD) and CDMA- Time Division Duplex (TDD). FDD mode is considered the main technology for UMTS. Separate 5 MHz carrier frequencies are used for the uplink and downlink respectively, allowing an end-user data rate up to 384 Kbits/sec. The TDD mode is Time-DivisionSynchronous Code-Division Multiple-Access (TD-SCDMA). TD-SCDMA operates on low-chip-rate carriers and allows end-user data rates up to 2Mbits/sec. 2.3 3G Network Architecture A 3G network consists of three interacting domains- a Core Network (CN), Radio Access Network (RAN) and the User Equipment (UE) also called a 3G mobile station. 3G operation utilizes two standard suites: UMTS and CDMA2000 which have minor differences with respect to the components they have in the RAN and the CN. The main function of the 3G core network is to provide switching, routing and transit for user traffic. It also contains the databases and the network management functions. The core network is divided into Circuit-switched (CS) and Packet-switched (PS) domains. Circuit switched elements include Mobile Services Switching Center (MSC), Visitor Location Register (VLR), and gateway MSC. These circuit switched entities are common to both the UMTS as well as the CDMA2000 standards.
3G Radio Access Ne twork 3G Core Network
Packet switched domain BS RNC BS UTRAN

SGSN PDSN

GGSN IP Network

UE
BS PCF

MSC HLR

GMSC

Circuit switched network

BS

cdma2000 RAN Circuit switched domain UMTS cdma2000 umts and cdma2000 BS: Base station; UE: User equipment; RNC: Radio network controller SGSN: Serving GPRS support node; GGSN: Gateway GPRS support node PDSN: P acket data serving node; MSC: Mobile switching center GMSC: Gateway mobile switching center; HLR: Home location register RAN: Radio Access Network UTRAN: UMTS Terrestrial RAN

Figure 1: Components of a 3G Network The differences in the CN with respect to the two standards lie in the PS domain. Packet-switched elements in UMTS include Serving GPRS Support Node (SGSN) and Gateway GPRS Support Node (GGSN). CDMA2000 packet-switched component is primarily the Packet Data Serving Node (PDSN). Some network elements like Equipment Identity Register (EIR), Home Location Register (HLR) are shared by both domains. The main function of the MSC server is to handle call-control for circuit-based services including bearer services, etc. The MSC server also provides mobility management, connection management and capabilities for mobile multimedia as well as generation of charging information. It can also be co-located with the Visitor Location Register (VLR). GGSN is the gateway to external data networks. It supports control signaling towards external IP networks for authentication and IP-address allocation, and mobility within the mobile network. GGSN provides functions for forwarding and handling user information (IP packets) to and from external networks (Internet/ intranets). SGSN provides session management, i.e. mechanisms for establishment, maintenance and release of end user Packet Data Protocol (PDP) contexts. It also provides mobility management and supports inter-system handoff between mobile networks. SGSN also supports generation of charging information. The PDSN incorporates numerous functions within one node. Routing packets to the IP network, assignment of dynamic IP-addresses and maintaining point-to-point protocol (PPP) sessions are some of its main functions. It also initiates the authentication, authorization and accounting (AAA) for the mobile station. The radio access network provides the air interface access method for the user equipment. An UMTS RAN (UTRAN) consists of Radio Network Controllers (RNC) and Base Stations (BS) or NodeB. The RNCs manage several concurrent Radio Link Protocol (RLP) sessions with the User Equipments

and per-link bandwidth management. It administers the Node-B for congestion control and loading. It also executes admission control and channel code allocation for new radio links to be established by the NodeB. A CDMA2000 RAN consists of a base station and 2 logical components- the Packet Control Function (PCF) and the Radio Resources Control (RRC). The primary function of the PCF is to establish, maintain and terminate connections to the PDSN. The PCF communicates with the RRC to request and manage radio resources in order to relay packets to and from the mobile station. It also collects accounting information and forwards it to the PDSN. RRC supports authentication and authorization of the mobile station for radio access and supports air interface encryption to the mobile station.

3. Related Work and Contributions

A lot of recent works have focused on the design and evaluation of architectures to integrate 3G and WLAN networks. Buddhikot et al. [9] described the implementation of a loosely coupled integrated network that provides roaming between 3G and WLAN networks. Tsao et al. [10] presented yet another method to support roaming between 3G and WLANs by introducing a new node called the virtual GPRS support node in between the WLAN and the UMTS networks. Their approach showed a reduction in handoff latency compared to the Mobile IP approach. Tsao et al. [12] evaluated three different internetworking strategies: the mobile IP approach, the gateway approach and the emulator approach with respect to their handoff latencies. Bing et al. [16] discussed mobile IP based vertical handoff management and its performance with respect to signaling cost and handoff latency. Matusz et al. [17] analyzed the performance of vertical handoffs between IEEE 802.11b and UMTS in the case of Mobile IPv4. They concluded that handoff packet delay and throughput depends upon the number of WLAN users and the traffic generated by them. All of the above works have focused on evaluating integration architectures in terms of the handoff latency experienced by the users when trying to move across 3G and WLAN networks while accessing a public network such as the Internet. In contrast to the above related efforts, our work mainly focuses on the endto-end delay experienced when users located in two different networks, namely the IEEE 802.11b and the UMTS communicate with each other directly. We describe two basic integration architectures that are used to connect these networks. Through simulations, the end-to-end packet latencies experienced by users when data is exchanged between these networks through the SGSN and the GGSN nodes are recorded and verified using various types of applications. The results demonstrate the feasibility of these integration architectures.

4. WLAN and 3G Cellular Data Network Integration Architectures- A Review

Table 1 reviews and summarizes the various 3G-WLAN internetworking strategies and their features. The Mobile IP [12] internetworking architecture considers 3G and WLANs as peer networks. It allows easy deployment but suffers from long handoff latency and might not be able to support real-time services and applications. The gateway approach [10] permits independent operation of the two networks and provides seamless roaming facility between them. It connects the two networks using a new logical node called the virtual GPRS support node. The gateway approach does not require the use of Mobile IP and has a comparatively lower packet loss during handoff. The emulator approach [12] is difficult to deploy since it requires combined ownership of the two networks but does yield low handoff latency and is thus well suited for real-time applications. Tight coupling [9] deploys WLAN as an alternative radio access network and uses a gateway between the WLAN and the 3G core network. It offers faster handoffs and high security mechanisms but also requires combined ownership of the two networks. Loose coupling [9] has low investment costs and permits independent deployment and traffic engineering of WLAN and 3G networks. However it suffers from high handoff delays and service disruption periods. Finally, the peer-networks approach allows easy deployment but also suffers from high handoff delays, thereby making it unsuitable for real-time applications. The choice of the integration architecture is important since multiple integration points exist with different cost-performance benefits for different scenarios.

Internetworking Approach

Complexity of deployment

Handoff Latency

Network Management and Ownership Separate ownership of 3G and WLAN networks permitted with roaming agreements Permits 3G and WLAN networks to operate independently.

Authentication

Billing

Real time application support

Mobility scheme

Mobile-IP approach

Easiest deployment

High

Reuse 3G cellular AAA functions in WLANs

Uses 3G billing feature

No

Use of Mobile IP

Gateway approach

Medium level deployment difficulty.

Low

Can provide both, separate WLAN security as well as 3G AAA.

Can provide separate as well as combined billing mechanism.

Yes

Emulator approach

Very difficult deployment

Low

Combined ownership required

Uses 3G authentication mechanism

Uses 3G billing mechanism

Yes

Intersystem roaming agreements between the networks through the use of a gateway. Mobility handled by UMTS Session management and GPRS Mobilitry Management. GPRS mobility management procedure used. Integration of Mobile-IP functionality in WLAN gateway. Mobile IP scheme used.

Tight coupling

Medium level difficulty

Low

Loose coupling

High level of difficulty

Higher compared to tight coupling

Peer networks approach

Very easy deployment

High

Generally requires both WLAN and 3G networks to be owned by same operator Permits independent deployment of WLAN and 3G networks. Both, same or different operator ownership permitted

3G ciphering key used for WLAN encryption Cellular access gateway provides authentication Use of AAA functionality of the 3G UMTS network

3G accounting features reused Billing mediator to provide common accounting Billing feature of 3G UMTS network used

Yes

Not very suitable

Not very suitable

Table 1: Comparison of various 3G-WLAN Internetworking Strategies

5. Simulated Architectures
We evaluated via simulations using OPNET two internetworking architectures to interoperate the 3G (UMTS) and WLAN networks by connecting them at two strategic points- the SGSN node and the GGSN node as shown in figure 2.
UMTSCore Network UMTS RAN RNC SGSN GGSN Internet

UMTSCore Network UMTS RAN RNC SGSN GGSN Internet

UE BS
UE: User Equipment AP: Access Point MN: Mobile Node BS: Base station SGSN: Serving GPRS support node GGSN: Gateway GPRS support node VGSN: Virtual GPRS support node RNC: Radio network controller GW: Gateway

UE BS AP
Flow of data AP MN Wireless LAN
UE: User Equipment AP: Access Point MN: Mobile Node BS: Base station SGSN: Serving GPRS support node GGSN: Gateway GPRS support node RNC: Radio network controller

MN

Wireless LAN
Flow of data

Figure 2: 3G-WLAN Integration - a) SGSN; b) GGSN

5.1 UMTS-WLAN Integration at the SGSN node When the UMTS and WLAN networks are connected through the SGSN node (figure 2a), the WLAN network does not appear to the UMTS core network as an external packet data network. Instead, it simply appears as another radio access network . The WLAN AP in this case needs to have the capability of processing UMTS messages. This WLAN access point possesses the features of a 3G (UMTS) radio network controller. On one of the interfaces this access point supports the functions of a RNC and on the other interface it communicates with the WLAN Mobile Node (MN) present in the WLAN. Thus, whenever a Mobile Node in the WLAN network wants to exchange data with the UMTS User Equipment, it first needs to undergo the GMM attach procedure (figure 3). This procedure is required to notify the SGSN of the location on the communicating node and also to establish a packet-switched signaling connection. The WLAN AP is responsible for sending these request messages to the SGSN on behalf of the WLAN Mobile Node. The GMM attach procedure is a three-way handshake between the communicating node, the RNC and the SGSN. Upon completion of this procedure, the WLAN Mobile Node is authenticated into the UMTS network. The WLAN access point in this case is aware of the UMTS network and sends messages from the WLAN Mobile Node to the UMTS User Equipment by converting each packet into a UMTS format by adding the appropriate packet data protocol (PDP) headers.
UE RNC/Specialized WLAN AP 1: GMM Attach Request SGSN

2: GMM Attach Request

4:GMM Attach Accept 5:GMM Attach Complete

3: GMM Attach Accept

6: GMM Attach Complete

7: PS Signaling Connection

Figure 3: GMM Attach Procedure 5.2 UMTS-WLAN Integration at the GGSN node In this type of integration (figure 2b), whenever a Mobile Node in a WLAN network wants to communicate with a User Equipment in the UMTS network, it does so through the GGSN node. The User Equipment in the UMTS network first activates the Packet Data Protocol (PDP) context that it wants to use. This operation makes the User Equipment known to its GGSN and to the external data networks, in this case, the WLAN network. User data is transferred transparently between the User Equipment and the WLAN network with a method known as encapsulation and tunneling. Data packets are equipped with PS-specific protocol information and transferred between the User Equipment and the Mobile Node in the WLAN through the GGSN node. The protocol that takes care of this is the GPRS Tunneling Protocol (GTP). It tunnels user data between the GGSNs and the SGSNs. For this kind of internetworking configuration, the WLAN AP is a simple 802.11b access point and does not require any special capabilities to process UMTS messages. It simply forwards the IP packets from the Mobile Node to the GGSN node. The packets are tunneled by the GGSN and delivered to the UMTS RAN. 5.3 Simulation testbed Application FTP FTP GSM encoded voice GSM encoded voice HTTP Web browsing QoS Class Background Background Conversational Conversational Interactive Measurement Parameter (seconds) File download time File upload time End-to-end delay Jitter Page response time Size 100-1000 Kilobytes 100-1000 Kilobytes 33 Bytes 33 Bytes 3000 Bytes Protocol TCP TCP UDP UDP TCP

Table 2: Description of various applications tested

A network simulation model was constructed using OPNET 10.0.A [18]. OPNET is a discrete event simulator with a sophisticated software package capable of supporting simulation and performance evaluation of communication networks and distributed systems. The simulation environment we used had a UMTS network connected to a WLAN network. The UMTS network was composed of the radio access network with a BS (Node-B), RNC, and User Equipment and a packet-based core network with SGSN and GGSN nodes. In order to simplify the simulation process, only one UMTS cell was used. The WLAN network is composed of 802.11b wireless Mobile Nodes configured in IBSS (infrastructure basic service set) mode. In the GGSN integration case, a simple WLAN access point was used, while in the SGSN integration case, a different access point with additional capability of processing UMTS messages was employed. The simulation used a standard OPNET 802.11b PHY module with maximum data rate up to 11 Mbits/sec to simulate the wireless medium. The goal of the simulations was to compare the delays involved when user data is exchanged between the UMTS and WLAN networks connected via two methods, namely GGSN and SGSN. We assume no congestion or queuing delays occur at any of the other intermediate nodes. Data is exchanged between a single User Equipment and a WLAN Mobile Node for a duration of one hour. Different types of traffic was generated using four different applications including Voice over IP (VoIP) in GSM encoded format, FTP, and HTTP (web browsing) as shown in table 2. These applications correspond to the various UMTS QoS classes. Conversational class is intended to carry real time traffic flows such as VoIP. Interactive and background classes are meant for bursty packet data transmissions and include traditional Internet applications such as FTP, HTTP etc. Interactive class is more suitable for web browsing where quicker response times are expected as compared to FTP. Packet delay, jitter, upload, and download response times were measured. Other parameters associated with each application are summarized in table 2. FTP flows are configured with file sizes in the range of 100 to 1000 Kilobytes with an increment size of 100 Kilobytes. Real-time VoIP flows were configured with a constant packet size of 33 bytes. Page size used for HTTP transfers was 3000 bytes.

6. Simulation Results and Discussion

Simulations performed for both UDP and TCP flows are presented. For the UDP flow (VoIP traffic), endto-end packet delays and jitter were measured. For TCP flows (FTP,HTTP) the upload/download response times were measured.
FTP: File Dow nload Response Tim e
8

FTP: File upload Response Tim e 8

GGSN SGSN

Download time (seconds)

7 6 5 4 3 2 1 0

SGSN

Upload Time (seconds)

GGSN

7 6 5 4 3 2 1 0

1 00

200 300 400 500 600 700 800 900 1 000

1 00 200 300 400 500 600 700 800 900 1 000

File Size (KB)

File Size (KB)

Figure 4: a) Download time; b) Upload time.

Voice Packet End-to-end delay End-to-end delay (seconds) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Voice Jitter 0.3

GGSN

GGSN SGSN

0.25 Jitter (seconds) 0.2 0.15 0.1 0.05 0

10 0 70 0

SGSN

19 00

31 00

13 00

31 00

19 00

25 00

37 00

Sim ulation Run-Tim e (seconds)

Sim ulation Run-Tim e (seconds)

Figure 5: a) End-to-end delay; b) Jitter. Figures 4a and 4b show the simulation run-times corresponding to the average file download and upload times experienced when transferring files of various sizes between the User Equipment and the WLAN Mobile Node under two different integration scenarios. We can see that the download/upload times are much lower for all file sizes when the transfer is done through the GGSN node as compared to when the networks are interconnected at the SGSN. In figures 5a and 5b the average delay and jitter for voice is presented. It is observed that both, delay and jitter values are much lower in the GGSN case. The average end-to-end delay is approximately 0.05 second (50 ms) in the GGSN case which is within the range of tolerable delay (0 to 150 ms) [20] while delay for the SGSN case is much higher. Similarly, figure 6 shows the response time to access a web page of size 3000 bytes. As figure 6 illustrates, the page response time is initially high and then decreases as the simulation progresses. We speculate that this reduction in the page response time may be because the web servers cache is initially empty and the first few page requests will cause the page to be fetched from the disk resulting in a high response time. As the more requests are generated with time, the cache is being filled and there is an increasing probability that one or more requests can be satisfied by the cache thereby reducing the overall page response time.
HTTP Page Response Tim e 35
GGSN

Page response time (seconds)

30 25 20 15 10 5 0
70 0 10 0

SGSN

13 00

25 00

31 00

19 00

Sim ulation Run-Tim e (seconds)

Figure 6: Page response time. The simulation results reveal that the application response time (delay) is consistently higher in the case where the UMTS and WLAN networks are connected through the SGSN node, as compared to the case where the two networks are connected at the GGSN. This higher response time can be attributed to the

37 00

13 00

25 00

37 00

10 0

70 0

additional processing time required at the WLAN access point in the first case. When the two networks are connected at the SGSN node, the WLAN access point performs the functions of a RNC on one interface and on the other interface implements the functionality of a WLAN AP. Therefore, it has to perform the additional initialization steps to authenticate the WLAN Mobile Node to the UMTS network, in the form of GMM Attach procedure and PDP context activations discussed in figure 4. Furthermore, internetworking through the SGSN requires the WLAN access point to be UMTS-aware and process each packet from the WLAN Mobile Node by adding the required packet data protocol (PDP) headers before it can be sent to the SGSN. This adds additional overhead to each packet. When integration is done at the GGSN node, the WLAN AP is a simple 802.11b access point and does not require any special capabilities to process UMTS messages. Data packets are transferred between the User Equipment and the WLAN network using encapsulation by the GPRS tunneling protocol. This reduces the packet latency as there is no additional delay due UMTS initialization procedures or packet conversions. The advantages, however, of using SGSN integration scheme include the reuse of UMTS authentication, authorization, accounting (AAA) mechanisms, usage of common subscriber databases and billing systems, increased security features (since the UMTS security mechanisms are reused), as well as possibility of having continuous sessions as users move across the two networks, since the handoff in this case is very similar to an intra-UMTS handoff as the WLAN AP appears as another RNC to the SGSN node. In the case of GGSN integration, since the WLAN is considered to be an external network, different billing and security mechanisms are needed. Service disruption is also possible during a handoff from one network to another.

7. Conclusions and Future Work

In this paper we discussed the architecture and performance of a 3G-WLAN integrated system connected at two different points namely the SGSN and GGSN. The architectures were evaluated with respect to the end-to-end latency, jitter and download/upload times obtained when data is exchanged between nodes located in the UMTS and WLAN networks respectively. Our simulation results show that the overall delays are much lower when the data exchange is done through the GGSN node as compared to when the networks are connected through the SGSN node. Additional delay in the latter case is attributed to the extra processing required at the WLAN AP which needs to perform the functionality of both a RNC as well as an 802.11b AP. Traffic passage through the GGSN is faster due to simple encapsulation procedure employed by the GPRS tunneling protocol. However, SGSN integration has its own advantages of providing strong security, common billing, authentication, etc. which are not possible when networks are coupled at the GGSN. Our future work focuses on evaluating these integration schemes with respect to the handoff latency and the development of an architecture that provides seamless session mobility when Mobile Nodes move across the 3G and WLAN networks.

References
[1] S. Jun-Zhao, J. Sauvola,, D. Howie, Features in Future: 4G Visions from a Technical Perspective, IEEE GLOBECOM, November 2001, volume 6, pages 3533 3537. [2] U. Varshney, R. Jain, Issues in Emerging 4G Wireless Networks, Computer Communications, June 2001, volume 34, Issue 6, pages 94- 96. [3] A. Miah, K. Tan, An Overview of 3G Mobile Network Infrastructure, IEEE Student Conference on Research and Development, July 2002, pages 228-232. [4] IEEE 802.11b, Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications- Higher-Speed Physical Layer Extension in the 2.4 GHz Band, 1999.

[5] U. Varshney, The Status and Future of 802.11- Based WLANs, Computer Communications, June 2003, volume 36, Issue 6, pages 102- 105. [6] J.D. Vriendt, P. Laine, C. Lerouge, X. Xiaofeng, Mobile Network Evolution: A Revolution on the Move, IEEE Communications Magazine, April 2002, volume 40, Issue 4, pages 104-111. [7] H. Luo, Z Jiang, B. J. Kim, N.K. Shankar, P. Henry, Internet Roaming: A WLAN/3G Integration System for Enterprises, AT&T Labs- Research, www.nd.edu/~hluo/publications/SPIE02.pdf. [8] Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications, ANSI/IEEE Std 802.11 1999 (E) Part 11, ISO/IEC 880211. [9] M. Buddhikot, G. Chandranmenon, S. Han, Y. Lee, S. Miller, Design and Implementation of a WLAN/CDMA2000 Internetworking Architecture, IEEE Communications Magazine, November 2003, pages 90-100. [10] S. Tsao, C. Lin, VSGN: A Gateway Approach to Interconnect UMTS/WLAN Networks, The 13th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, PIMRC 2002., volume 1, pages 275-279. [11] M. Jaseemuddin, An Architecture for Integrating UMTS and 802.11 WLAN Networks, 8th IEEE International Symposium on Computers and Communications, ISSC 2003, pages 716-723. [12] S. Tsao, C. Lin, Design and Evaluation of UMTS-WLAN Internetworking Strategies, Proceedings of IEEE Vehicular Technology Conference, VTC 2002, volume 2, pages 777-781. [13] A. Campbell. J. Gomez, S. Kim, A. Valko, C. Wan, Z. Turanyi, Design, Implementation and Evaluation of Cellular IP, IEEE Personal Communications, August 2000, volume 7, No. 4, pages 42-49. [14] A. Helmy, M. Jaseemuddin, Efficient Micro-mobility using Intra-domain Multicast based Mechanisms (M&M), August 2001 USC-CS-TR-01-747. [15] A. Salkintsiz, C. Fors, R. Pazhyannur, WLAN-GPRS Integration for Next-Generation Mobile Data Networks, IEEE Wireless Communications, October 2002, volume 9, Issue 5, pages 112-124. [16] B. Hongyang, H. Chen, L. Jiang, Performance analysis of vertical handoff in a UMTS-WLAN integrated network, Proceedings of 14th IEEE conference on Personal, Indoor and Mobile Radio Communications, PIMRC, Sept. 2003, volume 1, pages 187-191. [17] P. Matusz, P. Machan, J. Wozniak, Analysis of profitability of inter-system handoffs between IEEE 802.11b and UMTS, Proceedings of 28th annual IEEE International conference on Local Computer Networks, LNC, October. 2003, pages 210-217. [18] http://www.opnet.com [19] TS 125 422 V6.0.0, Universal Mobile webapp.etsi.org/action%5CPU/ 20040120/ts_125422v060000p.pdf Telecommunications System,

[20] B. Goode, Voice over Internet Protocol, Proceedings of the IEEE, September 2002, volume 90, Issue 9, pages 1495-1517.