A New Vertical Handoff Scheme for UMTS-WLAN Integrated Network

Transcription

1 A New Vertical Handoff Schee for UMTS-WLAN Integrated Network Mehrdad Manaffar 1, Haidreza Bakhshi 2 1,2 Telecounication Departent, Engineering Faculty, Shahed University, Tehran, Iran. Eail: 1 Manaffar@shahed.ac.ir, 2 Bakhshi@shahed.ac.ir Abstract Designing of heterogeneous Networks involves several interworking issues, such as sealess handoff or roaing, obility anageent, security and charging probles. Aong these interworking ites, the issue of sealess handoff is of great iportance for guaranteeing service continuity and Quality of Service. In this paper, we propose a sealess handoff schee with iniu handoff delay and low packet loss during UMTS WLAN handoff. The proposed handoff schee separates Layer 2 and Layer 3 handoff processes with help of a new defined signaling essage between obile nodes, called WLAN/UMTS Zone Identification Message and perfors pre-registration and preauthentication processes before entering the coverage area of the new Network and before perforing layer 2 trigger. Key Words-- Interworking, Mobile IP, UMTS, Vertical handoff, WLAN, Interworking. I. INTRODUTION A nuber of wireless network systes have been proposed and deployed during the past few years. It is believed that ultiple standards will coexist in the sae environent for the future wireless counication syste. Providing sealess roaing between heterogeneous networks becoes ore and ore iportant in the future obile counication syste. A ultiode UMTS/WLAN terinal can access high bandwidth data services where WLAN coverage is offered, while accessing wide area networks using UMTS at other places. However, this approach alone will only allow liited ulti-access functionality. To ake the ulti-access solutions effective, several interworking issues, such as dual ode support of obile, sealess handoff or roaing, obility anageent, security and charging probles ust be considered. Aong these interworking ites, the issue of sealess handoff is of great iportance for guaranteeing service continuity and Quality of Service (QoS. Therefore, in this paper, we survey current standard techniques for Mobile IP based sealess handoff, and propose an efficient sealess handoff procedure that supports iniu handoff delay and low packet loss in the proposed architecture. With help of a new defined signaling essage between obile nodes, the proposed handoff schee perfors preregistration and pre-authentication processes before perforing Layer 2 trigger. Moreover, it uses packet buffering and forwarding functions in order to reduce packet loss during the handoff period. On the above basis, detailed signaling procedures are presented, together with syste requireents when a obile terinal oves fro UMTS to WLAN and vice-versa. Quality of Service etrics that was used to evaluate the perforance of the proposed handoff schee are Average handoff delay, Packet loss and total Signaling costs and these etrics calculated by an analytical ethod. Nuerical results show that the proposed schee perfors well with respect to handoff delay, and packet loss copared with conventional schees and signaling cost still reains in an acceptable value. This paper is organized as follows. In Section 2, a survey on Mobile IP based vertical handoff schees will be presented. In Section 3, we provide requireents details of the proposed handoff procedure together with UMTS WLAN interworking architecture and syste requireents. In Section 4, overall handoff procedure fro UMTS to WLAN and vice versa will be discussed in details. In Section 5, we will try to calculate the Quality of Service paraeters and ake perforance coparisons between proposed handoff schee and other conventional schees and provide nuerical results and finally in section 6 a conclusion will be presented. II. PREVIOUS WORKS MIP is the ost widely known obility anageent proposal and is the ost coon solution for offering sealess roaing to obile devices on the Internet [7]. Its basic operation is consisted of advertiseent, oveent detection and registration, and packet delivery processes. MIP was originally designed to operate at the Layer 3 (L3 only, without regard to the underlying link layer (L2. This

2 approach iplies a clear separation between L2 and L3 handoff functionality but ay lead to unacceptable handoff latencies. Indeed, the essages generated by the registration process need soe tie to propagate through the network, and the obile node (MN is unable to send or receive packets during that tie [6]. For these reasons, the IETF s obile IP workgroup has proposed so-called low latency handoff schees, which are distinguished as Pre- Registration and Post- Registration handoff. In preregistration handoff, the network assists the MN in perforing an L3 handoff before the L2 handoff is copleted [3]. In post-registration handoff, the L3 registration occurs after the L2 handoff has been copleted. This technique uses a bidirectional edge tunnel (BET to support low latency and low packet loss during the change of the L2 attachent point of the MN [8]. Fig.1. Different UMTS-WLAN Interworking Architecture. III. PROPOSED HANDOFF SHEME Pre-registration and post-registration handoff schees are difficult to apply to the UMTS-WLAN interworking architecture currently under consideration because their operation was originally based on the MIP and they deal with only the registration procedure [3]. However, in the event of handoff between heterogeneous networks, both the registration process and the authentication process ust be included in the entire handoff procedure [9]. Therefore, we use a new handoff procedure taking into account actual handoff situations and network requireents in an integrated UMTS/WLAN network [10]. In this section, we describe the network architectures presented in [7] which is suitable for UMTS-WLAN interworking and explain in detail the operation of the proposed sealess handoff schee and handoff requireents. A. Interworking Architecture The considered network architecture for UMTS LAN interworking is based on interworking network odels of 3GPP standards docuents [2,7]. In [2], several necessary network entities are introduced for UMTS WLAN interworking and WLAN interworking reference odels are presented. After due consideration of the network architectures presented in current 3GPP standards, we were able to design the network architecture for the proposed sealess handoff schee, as shown in Fig. 1. B. WLAN/UMTS Zone Identification Message (ZIM The proposed handoff schee uses a prediction algorith to decide when to switch fro one access technology to the other. Fig.2. UMTS cell with a WLAN coverage area inside. To do this, all the obile nodes in a UMTS cell with coverage of one or ore WLAN networks inside, on a regular basis broadcast a specific signal, known as WLAN/UMTS Zone Identification Message (ZIM, to notify the other neighbor nodes, network type they are connected to. In ZIM signal, obile nodes inside WLAN coverage area e.g. MN2, broadcasts SSID of the WLAN service set which it is connected to, its own coordinates received fro GPS, WLAN access point s coordinates and the power of the received signal fro WLAN access point at their own coordinate and obile nodes inside the UMTS cell and outside the WLAN coverage area e.g. MN3, just broadcast the network type they are connected to which is UMTS network. As far as Accurate location deterination is alost ipossible without external location deterination systes and because of existence of the low cost GPS devices nowadays, we suppose that obile nodes will have an accurate GPS syste on their user equipent boards. In Fig. 2 a UMTS cell with a WLAN coverage area inside, is showed. As it can be seen sees a obile node (MN1 have a packet switched connection with UMTS network and in its trajectory inside the UMTS cell he/she is going to enter the WLAN coverage area. In all the conventional handoff schees proposed in section 2 of this paper, MN1 will wait until it enters the WLAN coverage area and the received signal strength fro WLAN access point reaches to the Receiving Threshold (Sensitivity of the MN1 s receiver

3 and then start to handoff the connection for UMTS to WLAN network to use the bandwidth and high data rate on WLAN network. In proposed handoff schee MN1, before entering the WLAN coverage area, will receive a ZIM signal fro one of the nodes inside WLAN area and will start to perfor the handoff procedure before the strength of received signal fro WLAN access point reaches to the MN1 s receiving threshold.. Handoff Decision Algorith In this section we are going to discuss about the propagation odel and received power of ZIM signal in different distances fro the node broadcasts it. There are three different propagation odels: free space odel, tworay ground reflection odel and the shadowing odel. These odels are used to predict the received signal power of each packet. In the free space odel, the world is assued to be flat and a direct path exists between the transitter and receiver. The receiving power P Rx depends only on the transitted power P Tx, the gain of the transit and receive antennas (G t,g r, the wavelength λ, the distance between both nodes d and a syste-wide loss coefficient L: P PTxGtGrλ = 2 16π d L Rx, Fx 2 2 The shadowing odel is an iproved version of the free space odel. It considers the existence of direct and indirect rays between the two nodes and it introduces a crossover distance d 0 after which the reflecting rays ay destructively interfere with the direct ray and drastically reduce the field strength. The power is coputed as: d PRx, SH [ db ] = PRx, FS [ db ] n log( N(0, σ (2 d 0 n is called the path-loss exponent and σ the shadow deviation. This odel also introduces a crossover distance d 0 after which the reflecting rays ay destructively interfere with the direct ray and drastically reduce the field strength. λ is the wavelength, G t and G r are the transitting and receiving antenna gain and L is the syste loss. Positioning ourselves in the worst of the cases we have chosen a value of 4 for n and a value of 7 for σ, as we are working inside buildings. When MN1 receives ZIM signal it calculates the exact direct distance between MN2 and WLAN access point (d using the coordinates of these two nodes received in ZIM signal. Then, MN1 calculates the border of the WLAN coverage area or in other words the radius (R of the WLAN coverage area. This is possible by using the equations (1 and (2 and also considering the MN1 s receiving sensitivity. In this radius fro WLAN access point, received signal fro WLAN access network will reach to the MN1 s (1 receiver s sensitivity. Once MN1 calculated the radius of WLAN coverage area and considering that MN1 knows the direct distance between itself and WLAN access point using the coordinates received fro GPS, it is going to calculate the exact tie it is going to pass the WLAN border. This value will be calculated regularly according to the GPS location update frequency and considering the MN1 s velocity at the oent. The calculated value will be copared with a specific threshold value (T threshold. If it is equal or less than Threshold, then the handover process will be initiated by MN1 otherwise the obile node will wait to reach to this threshold. This threshold value will help the obile node to choose the approxiately best tie to start the handoff process by considering its velocity and position. The threshold value itself could be deterined adaptively according to the WLAN and UMTS cell traffic. IV. HANDOFF PROEDURE A. Vertical Handoff fro UMTS to WLAN In this section we are going to propose a sealess Vertical handoff procedure that can iniize handoff latency and packet loss during UMTS WLAN handoff. The proposed handoff could be applicable to the considered UMTS WLAN interworking architecture without difficulty and guarantees service continuity and iniu handoff delay throughout the heterogeneous UMTS-WLAN networks. The proposed handoff schee perfors preregistration and pre-authentication processes with help of a new defined signaling essage between obile nodes, before perforing Layer 2 trigger. At the beginning, the MN1 counicates with the UMTS network. So, a packet transission route is fored between the HA and the MN1 through the UMTS network entities. Here, the HA anages the current oa (IP UMTS assigned by the UMTS FA according to the MN s hoe IP address (IP MN. Before the Mobile Node (MN1 in the UMTS syste enters the handoff region, it receives a WLAN Zone Identification Message (ZIM fro on of the obile nodes in the WLAN syste. WLAN obile node sends Service Set Identifier (SSID of the WLAN AP which the MN1 is going to enter. By receiving this essage MN knows that after a while it will enter the handoff region. So, it decides the acceptance or rejection of the handoff attept into the WLAN access network. After the handoff decision and before entering the handoff region MN1 request WLAN authentication by transitting the pre-authentication request essage to GGSN/FA and the GGSN/FA relays it to the PDG/FA of the WLAN. Here, the GGSN/FA is the current FA of the MN and the PDG/FA is the next agent to which the MN will ove. Because the GGSN/FA is assued to record PDG/FA addresses corresponding to the WLAN SSIDs that the MN1 receives, it can deliver the request essage to the

4 PDG/FA without error. This essage contains the MN authentication inforation that is required for standard WLAN authentication [2]. By receiving pre-authentication request essage, PDG/FA begins the standard Extensible Authentication Protocol/Authentication and Key Agreeent (EAP/AKA or Extensible Authentication Protocol/Subscriber Identity Module (EAP/SIM procedure for WLAN authentication aong the MN, PDG/FA and AAA server [2]. At this point, instead of the WLAN access network, the PDG/FA relays WLAN authentication essages between the MN1 and AAA server although the original WLAN authentication is perfored through the WLAN access network [19]. If pre-authentication is copleted successfully according to the standard authentication process, the EAP success is transitted to the WLAN access network with authentication keying aterial. This WLAN access network eorizes the received EAP success inforation that corresponds to the preauthenticated MN1, in order to use it after the obile node is attached to itself. The EAP success is also delivered to the MN via the pre-authentication reply essage. When the MN1 receives EAP success essage fro its GGSN/FA, it sends the PDG/FA a pre-registration request essage for obile IP (MIP registration. By receiving preregistration request essage in PDG/FA, a new oa (IP WLAN, which will be used in the WLAN, is assigned which can be either a foreign agent oa or co-located oa. Although a new oa (IP WLAN was assigned to the MN but incoing data packets are still send to the MN1 s old IP address on the UMTS network. Thereafter, the nfa sends the HA the pre-reg. update request essage, which includes the new assigned oa (IP WLAN together with the MN s hoe IP address (IP MN. If the HA receives the pre-reg. update request, it inserts the received oa into the next oa entity that corresponds to the MN1 s hoe IP address and then sends back the pre-reg. update reply essage toward PDG/FA. If the PDG/FA receives it, the preregistration reply essage is transitted to the MN1 through the GGSN/FA. This essage infors the MN1 of the copletion of pre-registration with the new assigned oa, IP WLAN. After the copletion of pre-registration phase, MN1, in its trajectory toward the WLAN coverage area, waits until it enters the WLAN coverage area and the power of the received signal fro the WLAN access point reaches to the sensitivity threshold of the MN1 s receiver equipents. At this point obile node sends a essage to the HA to substitute its current oa (IP UMTS with next oa (IP WLAN. By receiving this essage HA starts to buffer data packets sent to the MN s old IP Address on the UMTS syste. After this Mobile node disconnects fro the UMTS network by using the Layer 2 detachent process based on 3GPP standard [20]. Thereafter, it perfors the WLAN attachent process by using standard WLAN association procedures [9]. Naely, the L2 handoff is carried out fro UMTS to WLAN. Using previously received EAP access network, WLAN access network can confir that the new Mobile node is the pre-authenticated user. Therefore, the MN1 is able to counicate via the WLAN access network iediately, without a further WLAN authentication process [19]. After the MN becoes connected to the WLAN access network, it sends the handoff copletion essage to the HA and then HA ceases the packet buffering and forwards the data packets buffered up to this tie toward the obile node on it s new assigned IP address on the WLAN network. Now, the entire handoff procedures are copleted and the packet transission route is fored through the WLAN access network. In Fig. 3 and 4 the detailed handoff procedures of the proposed schee are presented. Fig. 3. Vertical handoff procedure fro UMTS to WLAN in proposed schee A. Vertical Handoff fro WLAN to UMTS In vertical handoff fro WLAN to UMTS, the obile node e.g. MN4, in its trajectory toward the outside of the WLAN coverage area receives a ZIM signal fro one of the nodes inside the UMTS cell e.g. MN3. By receiving this essage obile node understands that he/she is going to exit fro WLAN coverage area after a few seconds.

5 delay of a essage of size S sent fro node x to node y via wired and wireless links are presented respectively. S T w ( S, d x y = d x y ( Lw ( d x y 1 Pt (3 B S T wl ( S = ( Lwl B wl w (4 A. Signaling osts Total signaling cost ( VHO is the essage traffic exchanged between network nodes during handoff. Since the authentication and layer2 handoff phases are identical in all obile IP based vertical handoff schees, the only difference is in registration phase. If we show the signaling cost for registration updates during a session with reg, then reg for all handoff schees could be coputed as follows: Fig. 4. Vertical Handoff procedure fro WLAN to UMTS in proposed schee Then the obile node will wait until the power of the WLAN signal reaches to the receiving threshold of its receiver and then it will start to perfor the handoff process before exiting the WLAN coverage area. The overall handoff procedure that the obile node uses to ove fro WLAN to UMTS is depicted in Fig. 4. This case observes the sae signaling procedure as in the case of handoff fro UMTS to WLAN, except for the pre-authentication phase. When the obile nodes hands off fro WLAN to UMTS, pre-authentication is perfored aong the MN, SGSN and HLR/Au based on the 3GPP authentication procedure [11]. Because the UMTS authentication is issued by the SGSN, the pre-authentication request essage ust be sent to the SGSN, and so the SGSN ust process the UMTS authentication and transit the pre-authentication reply essage to the obile node following successful authentication. V. PERFORMANE EVALUATION In this paper we try to use the basic idea of analytical ethod presented in [7,8] to evaluate the perforance of proposed handoff schee and copared it with conventional schees such as the MIP, Pre-registration and Post-registration handoff schees and also sealess handoff proposed in [7]. In equations 3 and 4 transission ( MIP N Re g = S r (4d MN FA 2d FA HA Re g (Pr e. = S r (4d MN ofa 2d nfa HA 4d N ofa nfa Re g ( Post = Sr (4d MN nfa 2dnFA HA 2d ofa nfa N Re g ( Sea. = S r (2d MN ofa 2d nfa HA 2d ofa nfa N Re g (Pr oposed = Sr (2d MN ofa 2d nfa HA 2d N MN ofa (6 (7 (8 (9 (10 Total signaling cost for all handoff schees could be obtained as follows: VHO = Re g. Auth. L2HO (11 Which Auth and L2HO is sae for all schees. For the proposed handoff schee, the total signaling cost could be obtained as follows: VHO (Pr oposed g. Auth. L2 S d N = Re HO r MN HA (11 Fig. 5 shows the coparison of signaling cost for registration update process. We can see that the preregistration handoff schee has the largest signaling cost to perfor registration update. This is because the preregistration handoff schee consues ore signaling for pre-registration operation with the nfa than the other schees [7]. Siilarly, the post-registration handoff schee shows larger signaling cost copared with the basic MIP and the sealess handoff schee presented in [7] since it also requires additional signaling procedures for post-

6 registration before the handoff copletion. On the other hand, MIP has saller signaling cost than the preregistration or post-registration handoff schees because it perfors a siple registration process without other signaling process before the handoff copletion. However, the sealess handoff schee in [7] has the lowest hops and signaling for registration update copared with the other schees, so it shows the best perforance in view of signaling cost. schees show nearly the sae handoff delay perforance. Aong the, the proposed handoff schee shows the sallest handoff delay perforance. TABLE I AVERAGE HANDOFF DELAY FOR DIFFERENT HANDOFF SHEMES Handoff schee Average Delay (Sec. Mobile IP Pre-Registration Post-Registration Sealess handoff Proposed handoff Fig. 5. Signaling ost versus obile nodes resident tie (T r for different handoff schees B. Average Handoff Delay Average handoff delay is defined as the period in which the obile node detaches fro his hoe access network and attaches to the new network. In proposed handoff schee the interval for which it is disconnected does not contain the authentication and registration processing tie and just includes the layer 2 handoff delay and transission delays after layer 2 handoff. The average handoff delay of different obile IP base handoff schees in registration phase could be obtained as follow: 1 TRe g. ( MIP = 4Twl 4Tw, d MN FA (12 2T ad 2Tw, d FA HA T Re g. (Pr e. = 4Twl 4Twl, d MN ofa T 4Tw, d FA HA 4Tw, d ofa nfa ( Post. 4T ( S 4T ( S, d Re g. = wl r wl r MN nfa T 2Tw ( S r, d MN FA 2Tw ( S r, d FA HA ( Sea. 2T ( S 2T ( S, d Re g. = wl r wl r MN ofa 2Tw ( S r, d FA HA 2Tw, dofa nfa (13 (14 (15 The average handoff delay for different approaches is listed in Table I. MIP has uch longer handoff delay than the other schees because it does not use efficient handoff technique for vertical handoff. On the other hand, the other. Total Handoff Tie This is the total tie required to obile node start the handoff procedure and finishes it. Handoff tie starts fro the tie obile node feel a layer 2 trigger or receives a layer 2 trigger essage fro network. In all conventional IP based handoff schees obile nodes enters the new network coverage area and after receiving a layer 2 essage fro network, starts to initiate a vertical handoff. But, in proposed handoff schee the obile node before entering the new Network coverage area initiate the handoff and then waits to feel a need for layer 2 trigger and then disconnects fro his hoe network and attaches to the new network. In Table II total handoff tie required for different obile IP based vertical handoff schees is presented. As it sees the proposed handoff schee has the lowest required handoff tie. TABLE II TOTAL HANDOFF TIME REQUIRED FOR DIFFERENT HANDOFF SHEMES Handoff schee Average Delay (sec. Mobile IP Pre-Registration Post-Registration Sealess handoff Proposed handoff D. Total Packet Loss during Vertical Handoff According to the vertical handoff algoriths real packet loss starts fro the tie obile node disconnected fro its hoe access network and attaches to a new network [7]. So, the period of packet loss during vertical handoff is equal to the total handoff delay in all conventional obile IP based handoff schees. The total packet loss could be obtained as follows: Packet _ loss = T loss λ N (16 d T loss is the total duration in which all downlink in-flight packets will be lost. T loss is equal to the total handoff delay calculated in pervious section for each handoff schees; λ d is downlink packet transission rate and N is the Average nuber of oveents during a session.

7 TABLE III TOTAL PAKET LOSS FOR DIFFERENT VALUES OF MOBILE NODE'S RESIDENT TIME Mobile IP Pre-Registration Post-Registration Sealess handoff Proposed handoff Table III shows the total lost packet versus obile nodes resident tie (T r. By increasing the resident tie of obile node during a session, total lost packet will decrease. According to the siulation results the proposed handoff schee achieves the best perforance because its duration of packet loss is shorter than that of the others. E. Buffer Size The buffer size requireents for all schees becoe the aount of total packet loss generated by one handoff. Naely, the buffer size required to prevent packet drop is the sae as the aount of packet loss in each handoff schee [7]. In Table IV required buffer size for different approaches are calculated for a single handoff. TABLE IV REQUIRED BUFFER SIZE FOR DIFFERENT HANDOFF SHEMES Handoff schee Buffer Size (Kbyte Mobile IP Pre-Registration Post-Registration Sealess handoff Proposed handoff VI. ONLUSION In this paper, we introduced a practical UMTS-WLAN interworking architecture based on 3GPP standards and proposed a sealess handoff ethod which guarantees low delay and low packet loss during UMTS WLAN handoff. To enable low handoff delay, the proposed handoff schee perfors both pre-registration and pre-authentication processes before LAYER 2 handoff. To achieve low packet loss, it forces the HA to perfor packet buffering and forwarding functions while anaging two oas during the handoff period. Applying these basic ideas, we have presented concrete signaling procedures and syste requireents that take into account the interworking architecture when the UE is handed off fro the UMTS to WLAN and vice-versa. As results of analysis, the proposed handoff operation perfors well with respect to signaling cost, handoff delay, packet loss, and required buffer size copared with the conventional schees. In conclusion, the proposed handoff schee is able to guarantee sealess handoff supporting QoS and service continuity for the UMTS WLAN interworking. Therefore, if the current UMTS WLAN interworking syste is odified a little according to handoff requireents, the proposed sealess handoff schee can be applied usefully to integrated UMTS/WLAN networks without difficulty. REFERENES [1] Qian Zhang et al., Efficient Mobility Manageent for Vertical Handoff between WWAN and WLAN, IEEE oun. Magazine, Vol. 41, No. 11, pp , Nov [2] 3GPP syste to Wireless Local Area Network (WLAN Interworking; Syste Description, Rel.6, 3GPP TS , v6.3.0, Dec [3] Vijay K. Vara et al., Mobility Manageent in Integrated UMTS/WLAN Networks, IEEE I 2003, Vol. 2, pp , May [4] S. Taso and. Lin, Design and Evaluation of UMTS WLAN Interworking Strategies, IEEE VT 2002 Fall, Vol. 2, pp , Sep [5] Kalle Ahavaara et al., Interworking Architecture Between 3GPP and WLAN Systes, IEEE oun. Magazine, Vol. 41, No. 11, pp , Nov [6] J. McNair and F. Zhu, Vertical Handoffs in Fourth-Generation Multi network Environents, IEEE Wireless oun., Vol. 11, pp. 8 15, June [7] H.H. hoi, O. Song, and D.H. ho, "Sealess Handoff Schee Based on Pre-registration and Pre-authentication for UMTS WLAN Interworking", Springer Wireless Personal ounications journal, Vol. 41, pp , [8] K. Pahlavan et al., Handoff in Hybrid Mobile Data Networks, IEEE Personal oun., Vol. 7, pp , Apr [9] H.H. hoi, O. Song, and D.H. ho, A sealess Handoff schee for UMTS WLAN Interworking, IEEE Globalco 2004, Vol. 3, pp , Dec [10] IEEE , Wireless LAN Mediu Access ontrol (MA and Physical Layer (PHY Specifications, Standard, IEEE, Aug [11] Requireents and Architectures for Interworking between HIPERLAN/2 and 3rd Generation ellular systes, ETSI TR , v1.1.1, Aug [12] K. El Malki et al., IETF draft, Low latency Handoff in Mobile IPv4, May [13] Rajeev Koodli, IETF draft, Fast Handovers for Mobile IPv6, Sep [14] John W. Floroiu et al., Sealess Handover in Terrestrial Radio Access Networks: A ase Study, IEEE oun. Magazine, Vol. 41, No. 11, pp , Nov [15] I-Wei Wu et al., A Sealess Handoff Approach of Mobile IP Protocol for Mobile Wireless Data Networks, IEEE Trans. on onsuer Electronics, Vol. 48, No. 2, pp , May [16] N. Badache, D. Tandjaoui, A Sealess Handoff Protocol for Hierarchical Mobile IPv4, Int.Workshop Mobile Wireless oun Network, Vol. 2, pp , Sep [17]. Perkins, IP Mobility Support, RF 2002, Oct [18]. Blondia, O. asals et al., Low Latency Handoff Mechaniss and Their Ipleentation in an IEEE Network, Proceedings of IT18, Teletraffic and Engineering Vol. 5b, pp , Berlin, Gerany, Sep [19] O. asals, Ll. erda, G. Willes,. Blondia, N. Van den Wijngaert, Perforance Evaluation of the Post-Registration Method, a Low Latency Handoff in MIPv4, IEEE I 2003, Vol.1, pp , May [20] G.M. Koien and T. Haslestad, Security Aspects of 3G-WLAN Interworking, IEEE oun. Magazine, Vol. 41, No. 11, pp , Nov