QoS Guaranteed Handover Scheme for Global Roaming in Heterogeneous Proxy Mobile IPv6 Networks

Transcription

1 QoS Guaranteed Handover Scheme for Global Roaming in Heterogeneous Proxy Mobile IPv6 Networks Kwangsub Go 1 Misun Kim 2 Kyujin Lee 3 and Youngsong Mun 3 1 KREONET Center KISTI Seoul KOREA 2 Administrative Management Division Anti-Corruption and Civil rights Commission Seoul KOREA 3 School of Computing Soongsil University Seoul KOREA Abstract - Mobility and quality of service (QoS) are becoming the more important issues in wireless communications. The traditional Internet service is expanding into new access media and applications. Since wireless communication services are accompanied by frequent handovers at remote sites scalable and fast handover has become a prerequisite for ubiquitous communication. In this paper the differentiated service (Diffserv) model is deployed in heterogeneous proxy mobile IPv6 (PMIPv6) networks to satisfy the QoS guaranteed service and fast handover requirements. The operational procedures for QoS guaranteed global roaming are presented. In addition QoS management and handover cost evaluation schemes based on a mobile host s movement scope are proposed. This paper analyzes the reduction in handover delay in a network-based localized mobility management framework. We propose and analyze a PMIPv6 optimized with a global mobile access gateway (G-MAG) which is a network-based entity to further improve the handover performance in terms of handover delay while maintaining minimal signaling overhead in the air interface among converged heterogeneous wireless networks. The handover signaling procedures with host-based MIPv6 are compared with network-based proxy MIPv6 (PMIPv6) and fast PMIPv6 assisted by G-MAG to show how much handover delay reduction can be achieved. Analytical results show that the handover delay is significantly reduced. Keywords: QoS PMIPv6 handover global roaming cost evaluation. 1. Introduction Traditional Internet service does not consider issues with host mobility and QoS such as transmission delay packet loss ratio and bandwidth. However providing secure and seamless mobility and guaranteed QoS become more critical issues in wireless mobile services. Furthermore the services that are now being used on mobile networks including Internet broadcasting teleconferencing interactive role playing and telemedicine tend to require a predefined QoS. In the near future wireless networks are going to fully integrate different wireless access technologies to enable their users to exploit advantages of these various technologies and satisfy the QoS requirements of new applications. Various wireless communication protocols developed for different purposes are integrating and converging to ubiquitous communications which request QoS-guaranteed fast and seamless roaming services. The increasing demands for ubiquitous and mobile computing will require the integration of various wireless access technologies such as WLAN (wireless LAN) 3GPP (3rd generation partnership project) 3GPP2 (3rd generation partnership project 2) and IEEE To visualize these access technologies wireless networks have been converging to the Internet Protocol (IP). For instance the mobile IPv4 (MIPv4) [1] and mobile IPv6 (MIPv6) [2] protocols were already standardized by IETF (Internet Engineering Task Force). In addition the IEEE e amendment [3] enhances IEEE with mobility support for users moving at vehicular speeds and the WiMAX (Worldwide Interoperability for Microwave Access) Forum has adopted IP mobility [4]. An IP level mobility protocol is needed to enable the users of these future wireless networks to roam freely between various access networks. Global roaming guaranteed QoS and vertical handover will be prevalent in the near future. The Internet protocol IPv4 has been slowly progressing toward IPv6 and during this period both protocols have become a part of the Internet service infrastructure. The existing wired core Internet architecture that is linked to wireless access networks is going to evolve into wireless Internet environments. A large number of different wireless access networks are linked using a multi-hop infrastructure including notebook PCs PDAs and small sensors and are being served actively. The MIP (mobile Internet protocol) has been developed to provide mobile Internet services based on host mobility. Although it is a stable and mature technology there are some obstacles to overcome to be widely deployed. It is too heavy to be implemented on small mobile devices and complicated message exchange procedures must be handled by the mobile host itself. As a network-based localized mobility protocol PMIPv6 has been proposed to overcome the problems with MIP s host-based mobility and long handover latency [5]. The goal of PMIPv6 is specifying a simple extension of MIPv6 that would support network-based mobility for IPv6 hosts while reusing the signaling and features of MIPv6. As an alternative to MIP PMIPv6 reuses the MIPv6 entities and concepts as much as possible but the mobility management procedures are carried by the network devices. Moreover the mobility infrastructure in the PMIPv6 domain can provide mobility to an MN operating in IPv4 IPv6 or dual mode even if the transport network is not an IPv4 or IPv6 network. As a result a new standard for supporting an IPv4 host and IPv4 network in PMIPv6 was released [6] and a new protocol for supporting MIPv6 hosts attached in PMIPv6 is actively being discussed[7]. PMIPv6 is

2 applicable to various networks but is inappropriate for global roaming applications. To solve the long handover latency of MIPv6 and localized movement in PMIPv6 we propose a new scheme that provides appropriate QoS and global roaming. This scheme adopts a QoS management server and global MAG. It extends the localized movement scope of PMIPv6 to heterogeneous global networks such as different s or international regions without considering the MN s location and movement. It also analyzes the host mobility costs based on the MN s movement scope. The rest of this paper is organized as follows. Section 2 describes the related works for reducing handover cost and providing QoS schemes in MIP networks. In section 3 we describe the architecture and functionality of PMIPv6 networks deploying Diffserv. The mobility management procedures for handover cost evaluation and supporting differentiated services are discussed. Section 4 presents the analytical results for Diffserv and the handover cost based on the MN s movement. Conclusions and remarks are presented in section Related Works In MIPv4 an MN is identified by its home address regardless of its current point of attachment in the network and the MN is associated with a care-of-address (CoA) when it is away from home [1]. This triangular routing causes significant delay that degrades the handover performance. It is important for the mobility management to support fast and seamless handover with negligible delay which enables active services without disruption. Thus improvements were made and incorporated in a newer version of MIP called mobile IPv6 (MIPv6) to overcome some of the drawbacks [2]. It has been discovered that mobility can be more efficiently handled if the mobility management is divided into global mobility management and localized mobility management. Extensions of MIPv6 such as hierarchical mobile IPv6 (HMIPv6) [8] and fast handover for mobile IPv6 (FMIPv6) [9] have been proposed by IETF for efficient localized mobility management. The main goal of these localized mobility management protocols is to reduce the handover delay by localizing the registration of an MN [8] so that seamless service continuity can be achieved during roaming across wireless networks. Handover latency is mainly the result of delays caused by the discovery configuration authentication and binding update procedures associated with a mobility event. Most of the recently proposed mobility management schemes have been host based that is the MN is directly involved in mobilityrelated signaling. Studies on handover cost evaluation to reduce these delays have been performed in various MIPv6 networks [10-16]. Handover anticipation based on layer 2 (L2) trigger information is used to reduce the registration delay [10-14]. M. Lopez et al. proposed a proactive handover scheme [10] that only considers the predictive mode. Both the predictive and reactive modes are considered to optimize the handover delay in PMIPv6 but the performance was analyzed separately for each mode [11-12]. To enhance the handover performance K. Lee [13] proposed a cross-layering mechanism combined with IEEE e networks. For a more accurate performance evaluation in FMIPv6 S. Ryu et al. proposed combining the two modes and considered the probability of predictive mode failure (PPMF) which is affected by the radius of a cell velocity of a mobile host and L2 triggering time [14]. To enhance the handover performance in PMIPv6 they proposed optimizing the authentication delay and predicting the optimized route in [15 16] respectively. Although PMIPv6 performs better than the typical host-based MIPv6 and its extensions in terms of handover performance [11 12] it has a long handover delay when an MN moves away from its local area. Mobile Internet traffic has recently increased at an exponential rate and the new class of applications increases the need for QoS guarantees which pose big challenges for the current wireless communication environments. The gap between QoS provisioning and demand has been significantly enlarged. Integrated services (Intserv) [17] and Diffserv [18] were proposed to solve the QoS problem in IP networks [19 20]. Most QoS-providing models in wired and wireless networks focused on the buffer scheduling mechanisms based on traffic types. Various QoS provisioning schemes have been proposed for mobile IP networks [21-24]. However these studies did not consider the MN s movement. In this paper we propose a new scheme that allows an MN to move wherever it wants while receiving a guaranteed QoS. A priority queue model is used to provide differentiated QoS in our model. Because the BE traffic has the lowest priority the highest and medium priorities are allocated to the EF class and AF class respectively. To enable a performance analysis of the priority queue we use the M/G/1 [25] queuing model. 3. QoS Guaranteed Global Roaming Model in Heterogeneous PMIPv6 Networks 3.1. Basic Operation of PMIPv6 PMIPv6 reuses the infrastructure of existing MIPv6 protocols but adopts new entities such as a mobile access gateway (MAG) and local mobility agent () for networkbased mobility management. The MAG monitors the MN s movement in access links and sends signaling messages to instead of the MN. The is an anchor point of the MN that assigns a home network prefix to the MN and performs the HA role in PMIPv6 domain. It also manages the MN s reachability state in the domain. Typically the MAG functionalities are embedded in an access router and the is stacked in the gateway. There is an IP tunnel for transferring signals and data packets between an MAG and. When an MN is attached to the PMIPv6 domain the MAG tries to acquire the MN s profile from the policy server. That profile contains the MN s ID and IP address configuration method for access link and may include the MN s home network address of IPv6. When the MAG acquires the profile of the MN it

3 sends a proxy binding update message to register the MN s location to the. After receiving the proxy binding update message the sends a proxy binding acknowledge message including the MN s home prefix information and creates a bi-directional tunnel to the MAG. It also manages a routing table for transferring data to the MN and maintains the MN s reachability. Then the MN sets up its IP address through a router advertisement procedure to get information about its home network prefix and address configuration method. When an receives external packets from the PMIPv6 domain for the MN it forwards them through the tunnel to the MAG which will eventually forward them to the MN QoS Guaranteed Handover Scheme for Global Roaming PMIPv6 provides mobility without the participation of the host but its mobility management scope is restricted within an domain. However the host needs inter- and inter-internet service provider () movement as well as global roaming. In this section we propose a roaming scheme which includes global roaming. Based on the MN s movement scope which might be intra- inter-s or inter- domains we analyze the QoS guaranteed handover costs. Fig. 1 illustrates the proposed network architecture for QoS-guaranteed fast global roaming in heterogeneous PMIPv6. In this model we adopt new entities: a QoS agent (QA) to support differentiated service and global MAG (G- MAG) for fast handover when an MN moves between domains. This scheme could provide QoS guaranteed services depending on a subscriber s service level agreement (SLA) in PMIPv6 networks. More details of the procedures for an MN s movement and SLA management are provided in [26 27]. We assume that an network is composed of various access infrastructures according to its QoS level and locality. An has a global QoS agent (GQA or QA) that acts as a global QoS manager like a bandwidth broker in a Diffserv network. Neighboring GQAs can communicate with each other to establish an inter-domain QoS association such as SLAs. Each subnet of an may have its own local QoS agent (LQA or QA) and HA for QoS and mobility management within its local area. Each LQA can manage the resources within its subnet and serves the MN with a service profile. Three MN movement scope cases are shown in Fig. 1. In the first two cases an MN moves around within the same domain while it moves into a neighboring domain in the other case. In this figure (1) indicates intra- movement which means that an MN attached to MAG1 moves into the new access point (AP) linked to MAG2 where both MAGs are managed by the same 1. Movement (2) represents inter- movement. This means that the MN attached to MAG2 moves into MAG3 where the MAGs are managed by different s but both of them belong to the same domain. Movement (3) shows inter- movement which means that an MN moves into another domain s access point. This movement is called global roaming which causes complicated procedures for handover. It causes long delays and QoS degradation. A global MAG (G-MAG) may solve this problem. It is geographically located on the border between domains and used to connect the domains with a security association (SA). Thus the G-MAG can manage the s AAAs and QAs of both domains. In addition it is able to transfer its profile and authentication information from one to the other during the pre-inter-domain handover. The G-MAG is able to estimate the MN s location and detect its movement by tracing the access point where the MN is attached. Thus it can properly predict the point of the inter-domain handover execution. If the MN s inter-domain handover is imminent the G-MAG performs a pre-interdomain handover in advance of the inter-domain handover between the previous MAG and new MAG while the MN is still connected to the G-MAG Performance Analysis This section describes a QoS-guaranteed handover cost analysis model that depends on an MN s movement scope. The handover cost of PMIPv6 C HO can be expressed as the sum of movement detection latency T MD proxy binding update latency T PBND and router advertisement latency T RA : C T T T HO MD PBND RA (1) T T T T PBND AUTH QoS CONF where T PBND can be expressed as the sum of authentication delay T AUTH QoS management delay T QoS and address configuration delay T CONF. Fig. 2 QoS-guaranteed fast handover cost analysis. Fig. 1 Network architecture for QoS-guaranteed fast global roaming. Fig. 2 shows the QoS-guaranteed handover cost analysis model based on Fig. 1 which shows all three MN movement scenarios. The notations (1) (2) and (3) represent intra-

4 inter- and inter- movements respectively. As shown in (1) of Fig. 1 when an MN moves to MAG2 from MAG1 MAG2 should register the MN to its. Since the handover procedure is performed by the MN s movement detection when an MN attaches to a new AP that is covered by a new MAG (MAG2) the AP sends L2 handover messages to the MAG2. This causes MN movement detection delay T MD. When MAG2 detects the MN s movement it sends authentication and QoS profile request messages for the MN to the AAA and QA servers respectively. After receiving the authentication and QoS acknowledgement messages for the above queries MAG2 sends a proxy binding update (PBU) message to its. If the can allocate the home network prefix to the MN it sends a proxy binding acknowledgement (PBA) message to MAG2. MAG2 sends a router advertisement (RA) message to the MN when it receives a PBA message from the. The intra- handover in PMIPv6 is now completed. In this case both MAGs (MAG1 and MAG2) are in the same region of 1 and the MN s registration procedures follow the conventional PMIPv6 protocol [5]. The QoS-guaranteed handover cost of intra- movement in PMIPv6 network described in expression (2): C T T T 2( t t t t ) t C P HO P HO P MD P PBNB P RA ra a q am mr P MD ra P RA ra mr P PBND P AUTH P QoS P CONF T t T t t T T T T 2( ta tq tam) T 2 t T 2 t T P AUTH a P QoS q P CONF 2 tam (2) can be where t ra is the movement detection delay t mr and t am are wireless and wired delays and t a and t q are authentication and QoS setup delays respectively. An inter- handover cost analysis is shown in (2) of Fig. 2. Since PMIPv6 is designed to manage local area movement if an MN is away from its current domain registration procedures for the MN should be carried out through its home network. This causes a long delay and QoS degradation. To reduce the handover delay through the MN s home network we propose the use of a pre-handover mechanism before detaching the current MAG (MAG2). When an MN receives a new router advertisement message with a network prefix that is different from its current one it sends a request to its current MAG (MAG2) to send its profile to the new MAG (MAG3) before detaching from MAG2. In this message the MN s profile is included for movement management without link disruption. Since MAG3 receives the MN s profile from MAG2 it does not need to request authentication and QoS properties from its home network entities. Thus we can significantly reduce the handover cost. When an MN moves to a new MAG that belongs to a new domain the proposed fast handover cost is represented by equation (3): INTER INTER INTER INTER CfP HO TfP MD TfP PBND TfP RA 2( tra ta tq tam) tmr 6 taml INTER INTER TfP MD tra TfP RA tra tmr (3) INTER INTER INTER INTER INTER TfP PBND TfP AUTH TfP QoS TfP CONF TfP PRE REG INTER INTER TfP AUTH 2( ta taml ) TfP QoS 2( tq taml ) TfP CONF 2( tam taml ) INTER TfP PRE REG tmr tra where t aml is the delay for the MN s profile transfer INTER T fp PRE REG between the two MAGs and pre-handover delay which occurs before the MN s movement to support for the MN s movement between s can be neglected. An inter- handover cost analysis is shown in (3) of Fig. 2. If it has to be executed in the conventional PMIPv6 schemes the MN suffers from a long service disruption because of the home registration of the MN. When the MN moves from MAG3 to MAG4 the inter- handover latency becomes too long to support seamless service continuity. To reduce this long inter- handover latency we propose a new scheme that adopts a network entity called G-MAG as shown in Fig. 1. The G- MAG is located between domains and is connected to both domains with a security association (SA). Upon predicting an imminent inter-domain handover for the MN it performs the inter-domain handover while the MN is connected to the previous MAG (MAG3) in the previous domain ( A). Since the G-MAG has dual connections between domains it can gather and manage the information of both domains. It can also transfer the MN s profile and authentication information from one to the other during the pre-inter-domain handover. The G- MAG is able to estimate the MN s location and movement direction by tracing the AP where the MN attaches. Consequently the G-MAG can properly predict the interdomain handover execution point. Thus because the new MAG (MAG4) receives the MN s profile from the G-MAG it does not need to request authentication and setup QoS properties from its home network. Therefore the proposed scheme can reduce the interdomain handover latency by avoiding the signaling to the MN s home network. When an MN moves to a new MAG that belongs to a new domain the handover cost of the proposed scheme is represented by equation (4): INTER NTER NTER NTER CfP HO TfP MD TfP PBND TfP RA 2( tra tam ta tq ) 6 tamg tmr INTER NTER T t T t t (4) fp MD ra fp RA ra mr INTER INTER INTER INTER fp PBND fp AUTH fp QoS fp CONF INTER INTER fp AUTH amg a fp QoS amg q INTER fp CONF 2( am amg ) T T T T T 2( t t ) T 2( t t ) T t t where t amg is the delay between MAGs that belong to

5 different domains. To show our model s handover cost effectiveness we present a handover cost analysis model for the conventional PMPv6 in Fig. 3. It shows a cost evaluation for inter- and inter- handovers in the conventional PMIPv6. As can be seen in this figure the intra- handover cost is identical to (1) of Fig. 2. Thus the intra- handover cost in the conventional PMIPv6 is the same as expression (1). However when an MN moves to 2 from its previous (1) the new MAG (MAG3) and 2 have no information about the MN if these events occur in conventional PMIPv6 networks. The delay associated with acquiring the MN s profile authentication and QoS information through the MN s home network should be taken into account in addition to the handover costs for inter- and inter- movements in the conventional PMIPv6. Fig. 3 Handover cost analysis in conventional PMIPv6. In Fig.3 (2) and (3) represent the inter- and inter- handover cost evaluations. When an MN moves into a new or new domain that does not belong to its previous network there are no benefits to PMIPv6 because it was developed as a local movement management scheme. Thus all of the mobility procedures have to be carried out like in MIPv6 networks. When a new MAG such as MAG2 or MAG3 detects an MN s movement it has to acquire the MN s profile authentication and QoS information through the MN s home network. After receiving the MN s profile from its home network the new MAG sends a PBU message to its (2 or 3). After receiving the PBA message from the new the new MAG sends an RA message to the MN and the inter- or inter- handover is completed. The handover costs for these two cases are the same because the handover procedures for inter- and inter- movements are identical. The handover cost is expressed by (5): INTER INTER INTER INTER CcP HO TcP MD TcP PBNB TcP RA 2( tra ta tq t ' t ' tam) 6 tah tmr a q (5) INTER INTER TcP MD tra TcP RA tra tmr INTER INTER INTER IINTER TcP PBND TcP AUTH TcP QoS TcP CONF 2( t t t ' t ' t ) 6 t a q a q am ah INTER INTER TcP AUTH 2( ta tah t ') TcP QoS 2( tq tah t ') a q INTER cp CONF 2( am ah) T t t where t ah t a and t q are the delays for the MN s registration authentication and QoS management through its home network respectively. 4. Cost Evaluation Results In this section we compare the handover costs of the conventional scheme and our proposed models in heterogeneous PMIPv6 networks by service types and the MN s movement scope. The parameters used in this handover cost analysis are defined as follows. Let s assume that a CN generates data packets destined for an MN at mean rate α and an MN moves from one subnet to the other at mean rate β. The packet to Mobility Ratio (PMR) ρ α/β is defined as the mean number of packets received by an MN from a CN per movement. The parameters l c and l d are defined as the average length of a control packet and data packet respectively. Then their ratio l can be defined as l c /l d. The cost of transmitting a control packet is given by the distance between a sender and a receiver. The cost of transmitting a data packet is l0 times greater than the average cost of processing a control packet at any host (and forwarding data packets at an HA). The average delay time and packet loss probability are very important factors for QoS. In Diffserv networks ingress edges classify the traffic by SLA. We assume three service levels: EF AF and BE. We also assume that an access router accommodates different types of data packets from numerous connections. To serve these three types of traffic we use 3 buffers in the output buffer module of each node. In this paper each node has an M/G/1 queuing model for evaluating the performance of prioritized packets. Packets are summed to arrive in the queue according to a Poisson process with mean rates λ 1 λ 2 and λ 3 for EF AF and BE packets respectively. The service times for packets from each traffic class follow exponential distributions with mean rates of 1/µ 1 1/µ 2 and 1/µ 3 for EF AF and BE packets respectively. The mean offered load of the EF AF and BE packets in the buffer are ρ 1 λ 1 /µ 1 ρ 2 λ 2 /µ 2 and ρ 3 λ 3 /µ 3 respectively. The packet scheduling at the buffer module is as follows. First the server visits an EF buffer. If packets exist in the EF buffer it serves them until the buffer is empty. Otherwise the server visits an AF buffer and serves the packets in that buffer. After the service is finished for the AF buffer BE packets are served. To compare our QoS-guaranteed model with the existing extensions of MIPv6 we use the wired and wireless experimental results in [11 12]. The values are t mr t am t hc 10 ms t a t q t a t q t ah t ac 20 ms t ra 2 ms t amg 6 ms and t aml 4 ms. The control packet size and data packet size are assumed to be 100 bytes and 1024 bytes respectively and the buffer size K is assumed to be 100. The traffic arrival rates are λ λ and λ under a work-load of 0.5. Figs. 4 5 and 6 show the handover costs according to service type and MN movement scope. Fig. 4 shows the intra-

6 handover costs of MIPv6 and PMIPv6. In this case we represent the cost of handover from the conventional PMIPv6 (cpmipv6) and proposed fast PMIPv6 (fpmipv6) as PMIPv6 because when an MN moves around in an domain the handover procedures are identical and thus the handover costs are the same. As can be seen the difference between the handover costs of MIPv6 and PMIPv6 is significant because of the MN s CoA registration in MIPv6 networks. It also shows the service types are a significant element influencing the handover costs. As shown in Fig. 4 the main factor for the handover cost is the service policy. PMIPv6 shows very good performance compared to MIPv6. In the PMIPv6 network the handover costs for the EF AF and BE services are 9.64 times smaller than MIPv6 s for all of the traffic. Fig. 5 Inter- handover costs in PMIPv6 networks. Fig. 4 Intra- handover costs in PMIPv6 networks. The inter- and inter- handover costs in PMIPv6 are shown in Fig. 5 and Fig. 6 respectively. In these figures we omit the MIPv6 costs because the results are the same as those of Fig. 4. Thus we compare the results of the conventional PMIPv6 and proposed fast PMIPv6 schemes. When an MN moves between s the handover costs of fpmipv6 are significantly lower than those of cpmipv6. This is because when an MN moves to a different in a cpmipv6 network additional handover procedures to acquire the MN s profile are needed as in MIPv6. All of the fpmipv6 service types show better performance than the AF service of cpmipv6. The handover costs of fmipv6 are 2.15 times smaller than those of cpmipv6 for all of the traffic. This shows that our scheme is effective for inter- movement. Fig. 6 Inter- handover costs. Fig. 6 shows the inter- handover costs. In this case we adopt a G-MAG to eliminate the MN authentication process and QoS setup delay involving its home network. The handover costs are shown for BE AF and EF traffic in the conventional and fast PMIPv6. Our scheme reduces the handover cost by 1.65 times compared to cpmipv6. The results show that fpmipv6 is effective under global roaming circumstances and could be adopted in future global networks that require an MN that is fast and QoS-guaranteed roaming.

7 5. Conclusions Various applications and portable communication devices demand various QoS and mobility levels under different circumstances. To provide a guaranteed QoS and fast mobility we propose procedures for mobility and QoS management in heterogeneous PMIPv6 networks. When Diffserv is deployed in a mobile IP network various problems are encountered. The most significant problem is related to the acquisition of the service profile for the MN when it moves into a different domain. Since the first-hop router does not have any information about the newly attached MN a guaranteed QoS could not be provided. To solve this problem we present procedures for acquiring the MN s service profile and additional information based on the MN s movement scope. Through a guaranteed-qos handover cost analysis and evaluation we reduce the handover cost by 9.64 times compare to MIPv6 for intra- movement. Moreover our scheme shows handover cost efficiencies for inter- and inter- movements that are 2.15 and 1.64 times better than the conventional PMIPv6 respectively. Consequently our scheme is effective under global roaming circumstances and could be adopted in future global networks that require a fast speed guaranteed QoS and roaming. In future work we will refine the message types and registration procedures to manage the MN s mobility more effectively and reduce the total communication cost. An authentication mechanism for seamless handover will also be considered to construct a service that is more efficient and secure with a guaranteed QoS in heterogeneous PMIPv6 networks. 6. References [1] C. Perkins IP mobility support for IPv4 IETF RFC 3344 Aug [2] D. Johnson C. Perkins and J. Arkko Mobility Support in IPv6 IETF RFC 3775 June [3] Part 16: air interface for fixed and mobile broadband wireless access systems-amendment 2: physical and medium access control layers for combined fixed and mobile operation in licensed bands IEEE Std e-2005 Feb [4] Stage 2: architecture tenets reference model and reference points. Part 2 WiMAX Forum Network Architecture Feb [5] S. Gundavelli K. Leung V. Devarapalli K. Chowdhury and B. Patil Proxy mobile IPv6 IETF RFC 5213 Aug [6] R. Wakikawa and S. Gundavelli IPv4 Support for Proxy Mobile IPv6 IETF RFC 5844 May [7] G. Giaretta Interaction between PMIPv6 and MIPv6: Scenarios and Related Issues draft-ietf-netlmm-mip-interactions-06 May [8] H. Soliman C. Castellucia K. Malki and L. Bellier Hierarchical Mobile IPv6 Mobility Management (HMIPv6) IETF RFC 4140 Aug [9] R. Koodli Fast Handovers for Mobile IPv6 IETF RFC 4068 July [10] A. Dutta S. Das D. Famolari Y. Ohba K. Taniuchi V. Fajardo R. M. Lopez T. Kodama and H. Schulzrinne Seamless proactive handover across heterogeneous access networks Wireless Personal Communication Vol. 43 Issue 3 pp Nov [11] C. Makaya and S. Pierre An Analytical Framework for Performance Evaluation of IPv6-Based Mobility Management Protocols IEEE Trans. on wireless communications Vol. 7 No. 3 pp March [12] K. Kong W. Lee Y. Han M. Shin and H. You Mobility management for All-IP mobile networks: mobile IPv6 vs. proxy mobile IPv6 IEEE Wireless Communications vol. 15 no. 2 pp [13] K. Lee and Y. Mun Enhanced Cross-Layering Mobile IPv6 Fast Handover over IEEE e Network will appear in IEICE Transaction on Communications. [14] S. Ryu K. Lee and Y. Mun Optimized fast handover scheme in Mobile IPv6 networks to support mobile users for cloud computing Journal of Supercomputing published in Online First DOI /s June [15] L. Magagula and H. A. Chan Optimized handover delay in Proxy Mobile IPv6 using IEEE MIH Services Military Communications Conference MILCOM IEEE pp. 1-7 Nov [16] J. Na S. Ryu K. Lee and Y. Mun Enhanced PMIPv6 Route Optimization Handover Using PFMIPv6 IEICE Trans. Commun. Vol. E93-B No. 11 pp Nov [17] R. Braden D. Clark and S. Shenker Integrated Services in the Internet Architecture: an Overview IETF RFC 1643 June [18] S. Blake D. Black M. Carlson E. Davies Z. Wang and W. Weiss. An Architecture for Differentiated Services IETF RFC 2475 Dec [19] M. May S. Blot A. Jean-Marie and C. Diot Simple Performance Models of Differentiated Services Schemes for the Internet Infocom 99 vol. 3 pp March [20] S. Bakiras and V. O. K. Li "Efficient Resource Management for End-to- End QoS Guarantees in Diffserv Networks 2002 IEEE International Conference on Communications Vol. 2 pp April [21] K. M. Yusof and N. Fisal "Provisioning QoS in Differentiated Service Domain for MIPv6" 4 th National Conference on Telecommunication Technology 2003 pp January [22] R. Jain T. Raleigh C. Graff and M. Reschinsky "Mobile Internet Access and QoS Guarantees using Mobile IP and RSVP with Location Registers" ICC98 pp June [23] A. Mahmoodian and G. Haring "Mobile RSVP with dynamic Resource Sharing" IEEE Wireless Communications and Networking Conference 2000 Vol. 2 pp September [24] R. Chakravorty M. D'Arienzo I. Pratt and J. Crowcroft. "A Framework for Dynamic SLA based QoS Control for UMTS" IEEE Wireless Communications Special Issue on Merging IP and Wireless Networks Vol. 10 No. 5 pp October [25] L. Kleinrock Queueing Systems vol.1. J. Wiley & sons [26] M. Kim S. Park and Y. Mun QoS Guaranteed Service Model in Mobile IPv6" SAM'03 Las Vegas U.S. vol. 2 pp June [27] M. Kim and Y. Mun Cost Evaluation of Differentiated QoS Model in Mobile IPv6 Networks" ICCSA 2006 Glasgow U.K. vol. 2 pp May 2006.