On End-to-End Mobility Management in 4G Heterogeneous Wireless ...

On End-to-End Mobility Management in 4G Heterogeneous Wireless Networks Muhammad Yousaf

Amir Qayyum

Center for Advanced Studies in Engineering (CASE), Islamabad [email protected]

M. A. Jinnah University, Islamabad aqayyumaieee.org

Abstract - With the advent of modern technology, mobile devices with multihomed capabilities are proliferating. Existence of different network interfaces in multihomed devices enables them to seamlessly roam across heterogeneous networks. These vertical handovers however, causes to affect the TCP connections. Many solutions exist that handle this problem caused by mobility of the nodes. However, current mobility management solutions either require support of additional entities in the network or require changes in current implementation of TCP. This paper argues that mobility management service can effectively be provided without requiring the support of additional entities in the network and without changing the current implementation of TCP. On the basis of this principle, we present an End-to-end Mobility management Framework (EMF) that overcomes the limitations of current mobility management solutions and provides a richer set of mobility services such as location updates, soft handover, willful handover, etc. Index Terms - Cross Layer Design, End-to-End Design, IEEE 802.21, Location Updates, Mobility Management, TCP, Vertical Handover, Willful Handover

In recent years, the technology scene as well as user's expectations from communication networks has changed enormously. It is now becoming common for a user to have access to a number of wired or wireless networks simultaneously. Moreover today's user does not want to be restricted withm a confined area while availing the communication facility. People desire to remain connected while on the move. Mobility management thus becomes one of the core challenges of future 4G heterogeneous all-IP networks. An important characteristic of heterogeneous networks is that the user equipment is multihomed. It is common that modern mobile terminals possess multiple network interfaces e.g. Ethernet, WLAN, Bluetooth and optionally GSWGPRS, WiMAX network interfaces. Currently, standards lack in efficiently exploiting the existence of these multiple network interfaces. Different network technologies have different geographical scope and coverage area. Due to its wired connectivity, Ethernet does not allow the users to move around but provides the higher data rates of lOOMbps and 1Gbps. IEEE 802.11 Wireless LAN provides mobility service in a limited locality and provides moderate data rates ranging from 5.5Mbps to 54Mbps. GPRSEDGE provides a larger coverage area but with much lower data rates ranging from

978-1-4244-2152-7/08/$25.00 02008 IEEE

9.05Kbps to 384Kbps. These different network technologies often have overlapping coverage areas, for example when users have EthernetIWLAN coverage, they normally are also in the coverage area of GPRSEDGE. Due to higher available bandwidth and lower cost users usually prefer to remain connected with local area network. When users move away from the coverage area of LANs, they might wish to continue their connections in the wireless cellular network in spite of lower available bandwidth and higher cost but with the larger coverage area. This migration of connections from one kind of access network to another kind of access network is termed as vertical handover. Applications running over TCP suffer from these handovers. Many solutions have been proposed to facilitate the vertical handovers. But most of these solutions require either the support of additional network entities i.e. home agent, foreign agent, mobility anchor points, proxy, etc or the changes in TCP itself. In this paper, we discuss a new End-to-End Mobility Management Framework (EMF) that overcomes both these problems of the existing mobility management schemes. Not only this new architecture does not require support from the network infrastructure but it also does not demand any change in the current implementation of TCP. Rest of the paper is organized as follows: Section 2 describes the related work. Section 3 discusses the design principles that served as the basis of EMF framework. Section 4 describes the EMF architecture. Section 5 explains the handover mechanism using EMF architecture. Section 6 describes the use of secure DNS dynamic updates for location management in EMF architecture, and in the end section 7 concludes this paper.

So far, various protocols have been developed to handle the mobility management of mobile nodes. These solutions can be categorized according to the layer of network protocol stack on which these solutions operate. One such categorization can be found in [I]. Almost all wireless access technologies e.g. IEEE 802.11 WLAN, WiMAX, GSWGPRS, WCDMAlCDMA2000, etc. have their own well-defined handover procedures [2], [3], [4]. These procedures are quite efficient when mobile nodes move from one cell to another cell of the same access technology. This is known as horizontal handover. For handovers from one type of access network to a different type of access network, upper layer solutions are usually incorporated. This kind of handover is known as vertical handover [5].

Majority of mobility management solutions revolve around Mobile-IP that is a network layer solution for mobility management. Variants of Mobile-IP (MobileIPv4 as well as MobileIPv6) provide solutions for both macro and micro mobility management [6], [7], [8], [9], [lo]. Although these are most mature among all the mobility management solutions, they have some limitations that are common to all the variants of network layer solutions. First, end-nodes using Mobile-IP require some supporting entities in the network to support the end-node's mobility; examples include home agent, foreign agent and Mobility Anchor Point (MAP) [ll]. This means that users would demand more functionality from the network. We argue that mobility is such a service that can be effectively provided without sacrificing the "smart-edges simple-network" theme of the Internet. Secondly, not all kind of mobility services can be implemented at the network layer. E.g. session mobility from one terminal to another terminal requires some upper layer processing that cannot be implemented at the network layer [12]. Thirdly, mobile-IP is not able to exploit multihoming for bandwidth aggregation by data striping [13]. Moreover Mobile IP also does not support the willful handovers. Many researchers believe in end-to-end handling of mobility management thus provide solutions at the transport layer of TCPIIP protocol stack. Examples of such solutions include TCP-Migrate [14], MSOCKS [IS], [16], pTCP [17] and SCTP [IS], [19], [20], TCP-Migrate, pTCp and MSOCKS require changes in the existing TCP. Moreover, for MSOCKS, in order to splice the TCP connection, an additional entity i.e. proxy is required in the network. SCTP, with its mSCTP and cSCTP extensions, has a good potential to cater the mobility management issues in an elegant manner. But the problem for applications running over TCP still remains unsolved. Solutions also exist that resolve these issues at the session layer. For example [21] presents a session layer mobility management (SLM) solution. However, SLM doesn't provide willful handovers. Also as there is no cross layer entity in the architecture thus SLM doesn't have provision to utilize the IEEE 802,21 MIH facility, Moreover SLM requires the support of a new additional entity User Location Server (ULS) in the network. Another limitation of SLM is that it is not able to exploit the presence of multiple network interfaces simultaneously thus bandwidth aggregation is not possible in SLM architecture. Session Initiation Protocol (SIP) also facilitates the mobility with its location updates option [22], [23]. SIP has gained popularity in the 3GPP community, but SIP too does not provide complete set of mobility services. SIP implementation of 3GPP is optimized for cellular networks thus some minor dfferences exist between IETF's SIP and 3GPP's SIP. In the heterogeneous 4G wireless networks, these differences can create interoperability issues [24]. SIP also requires some additional network entities e.g. Proxy Server, Registrar, etc. Moreover, SIP does not provide mobility solution for applications using TCP. Some cross-layer solutions have also been proposed for mobility management. A cross-layer manager is introduced in [25] that interact with multiple layers of the protocol stack to

create an interlayer coordination model. This idea is further extended in [26] in the form of Mobility Manager. This Mobility Manager utilizes IEEE 802.21 Meda Independent Handover (MIH) events, commands & information from the lower layers [27] and provides the handover commands to Mobile IPv6. Further more, Mobility Manager also provides notifications to upper layers for application adaptations. Both of these solutions use MIPv6 for mobility management that have inherent drawback of requiring number of network entities. MIPv6 also doesn't support data striping. 111. DESIGN PRINCIPLES OF EMF FRAMEWORK Initially Internet was designed on the principle of "smartedges simple-network model". The simple, flexible and scalable design along with deployment of low cost equipment enabled Internet to emerge as the blockbuster of the century. Earlier network designers strongly believed in the design that the services that can be easily implemented at the end hosts should be provided at the end hosts. At that time the major concern of Internet was to merely provide the connectivity. However with the advancements in computing as well as communication technologies, user expectations from the communication networks have changed to a large extent. Today's user demand services like mobility, quality of service, the Internet in the security, etc. that were not provided ages. In order to provide such services, now network designers are changing their minds and putting more and more intelligence in the network [28]. It emerged as a hot debate that whether the intelligence should be implemented in the network or in the end systems [29]. However, in spite of changing network requirements, many researchers still believe in the End-to-End design philosophy of the Internet [30]. Rather than waiting for the settlement of this "end-to-end or network intelligence debate", the framework proposed in this paper has been designed on the principle that if the implementation of network intelligence is present, then end systems should be able to utilize k s intelligence, but in the situations where network intelligence is not deployed, end systems should have sufficient intelligence to continue their service in a seamless way. Mobility management is such a service that can be implemented on k s principle. In order to develop an efficient and comprehensive mobility management framework, we draw following guidelines that served as design guidelines of the EMF framework: i) If network intelligence is present then end nodes should utilize k s intelligence for mobility management. However, if network support is not present then EMF framework should overcome this limitation and enable the mobile nodes to manage the mobility. ii) EMF framework should be designed in accordance to the end-to-end design philosophy of the Internet. It should not require any extra entity in the network to provide mobility service. iii) The objective of EMF framework is to facilitate the TCP rather than to change the current implementation of TCP.

iv) Applications using TCP, as reliable transport protocol, should continue having confidence of reliable data delivery for applications using the EMF framework. v) Current standards lack in providing the willful handovers. Therefore EMF framework should overcome this limitation and should support both forced handover as well as willful handover. vi) In order to provide the service continuation, EMF framework should support the soft handoffs. At the transport layer it can be provided in the form of connection diversity. vii) Mobile node should remain able to serve as mobile server. So, EMF framework should also provide the location management service. viii) Overhead regarding the handoff and location management should be minimized. ix) EMF framework should provide minimum handoff delays. x) Applications that do not require mobility support should not incur the mobility support overhead. xi) Appropriate security measures should be taken to avoid certain security attacks regardng the handoff and location update messages e.g. session hijacking attacks, denial of service attacks, etc. xii) Along with providing the vertical handoffs, EMF framework should also utilize the presence of multiple network interfaces for supporting other services e.g. data striping for bandwidth aggregation. IV. EMF ARCHITECTURE As described earlier, link layer solutions are not sufficient for the mobility across 4G heterogeneous IP networks. Network layer solutions have their own limitations and demands. Moreover it is not possible to provide complete set of mobility services at the network layer. Developing application specific solutions is also not an appropriate design choice. Hence, the appropriate implementation choice for our protocol is transport layer. EMF architecture is shown in the Figure 1. This architecture includes four main components: i) EMF Data Handler Protocol ii) EMF Control Protocol iii) EMF Host Agent iv) EMF User Agent A. EMF Data Handler and Control Protocols The core of the EMF architecture consists of two protocols: (i) EMF Control protocol and (ii) EMF Data Handler protocol. EMF Control protocol is used to exchange handover messages among communicating peer hosts. EMF Control is also responsible for sending the location updates using secure DNS dynamic update messages. While EMF Data Handler has the ability to simultaneously handle multiple data streams. EMF Control protocol establishes an association between two mobility aware applications. By "mobility aware" we mean that the end hosts have the implementation of EMF framework and the applications are using this framework.

Applications

EMF

userAgent

User

r---

%FZ&CG;S?

-----I +cornman&: Handover

Data Handler TCP

EMF ~~~t

I

j

i Agent ;

UDP

I I

I I

Events, %;mman~S?-

Information

Fig. 1. EMF Architecture

Each EMF association is identified by a unique identifier. A single EMF association can have multiple TCP connections. The emphasis is on the reliable delivery of user data rather than on the conservation of TCP connection. Due to network interface change or location change, if TCP connection is affected, then EMF Control establishes a new TCP connection with the same previous association identifier but with possibly new network layerltransport layer identifiers. It is the responsibility of EMF Data Handler to assure the in-sequence reliable transportation of user data. In [20], similar theme is used by SCTP-based solution that maintains multiple streams for a single association. Whenever an EMF compliant client requests a TCP connection, the EMF Core Protocol holds the connection request and checks whether the peer end is also EMF compliant or not by sending an EMF Association Initiation Request to the corresponding server. EMF control messages like Association Initiation Request and Association Change Request messages are of request-response type messages where no bulk data transfer is required. Therefore, these control messages can be exchanged over UDP. This design choice reduces the latency that would be involved in establishing the subsequent data connection. If client does not receive positive EMF response, it concludes that the peer end is not EMF compliant and proceeds with a normal TCP connection establishment for user's request. However, if peer end responds as EMF compliant, then both ends will agree on certain parameters. One such parameter is the unique EMF association ID. The client can send it in the Association Initiation Request message. To make sure the uniqueness, t h ~ s ID can be derived from the parameters like random number, client's IP address, timestamp, etc. The peer end should check that this suggested ID has not been used for any already established association. Although probability of having such match is rare but still if found then peer end can send its suggested unique association ID to the client node. In t h ~ s way, peer entities can agree on the unique ID. This ID needs to be unique only among mutually communicating parties and

need not be globally unique. Due to certain security concerns (e.g. session hijacking) some Public-Key cryptographic procedures may be adopted to assure the uniqueness, randomness and secure agreement of the association ID. For example, algorithms like Elliptic Curve Cryptosystems for shared key exchange may also be used to establish the shared association ID among peer nodes [3 11. After agreement on association parameters, client can now initiate the TCP connection as per user's request. In t h ~ s way an umbrella of EMF association is established beneath which multiple TCP connections can be maintained. There are two main objectives of this umbrella association. First is to hide the effects of mobility from the user application. On change of network interface address, EMF Control part initiates the handover mechanism and establishes a new TCP connection with the previous association ID but with new IP address. Details of this handover mechanism are discussed in the subsequent section. B. EMF Host Agent EMF Host Agent is the entity that collects local host information and statistics that are helpful for handover procedure. The primary objective of EMF Host Agent is to assist EMF Core Protocol for handovers. Two kinds of handover decisions are possible: i) Forced Handovers & ii) Willful Handovers. Forced handovers are handovers in which mobile terminal has no other choice except to handover its connection to the only available network. This kind of handovers is usually required when mobile terminal moves away from the network in which it has initiated the connection and enters in coverage area of another network. As previous network is no more accessible, thus in order to enjoy the service continuation, mobile terminal is forced to handover its connection to new network. These forced handovers are most often triggered due to changes in lower layer parameters. EMF Host Agent is responsible to monitor these changes of lower layer parameters. Events corresponding to these changes, for example, include access router discovery, obtaining new IP address or usage of IEEE 802.21 eventsltriggers. Willful-handovers are initiated, for example, when application or user preferences, obtained from the upper layers demand the connection migration. With multihomed mobile devices in heterogeneous 4G wireless environment, mobile users would have subscription to a number of network operators. If user has access to multiple networks simultaneously, then handover decisions will largely be affected by the user preferences. User preferences would depend upon parameters like available bandwidth, quality of operator's service, cost of connection, etc. This type of handover is called the Willful Handover. EMF Host Agent is required to interact with upper layers in order to get necessary information to initiate willful handovers. Current mature mobility management solutions, e.g. Mobile-IP variants, lack this facility of willful handovers. In t h ~ sregard, our proposed solution is comprehensive enough to incorporate both types of handovers i.e. forced as well as willful handovers.

As EMF Host Agent is required to interact with multiple layers of the network protocol stack, an efficient and optimized implementation design choice is the cross-layer solution. Figure 1 depicts the interaction of EMF Host Agent with corresponding layers of the protocol stack. At the lower layer, it interacts with the IEEE 802.21 MIH layer, and at the application layer, it interacts with the EMF User Agent to get user preferences. On the basis of this information, it interacts with the EMF Control module to assist the handover procedure. In its design, EMF Host Agent can be seen as cross-layer manager of [25] and mobility manager of [26]. The difference is that the cross-layer manager and mobility manager interact with Mobile-IPv6 as mobility management protocol, and EMF Host Agent interacts with the EMF Control module for the mobility management. C. EMF UserAgent To assist the willful handovers, we propose that there should be some application layer utility that can interact with the users. We call such utility as EMF User Agent. User can provide the details such as connection subscriptions, preferences, default settings, etc. in two ways: one is manual and the other is automated. It is likely that users will not be changing their network subscriptions and preferences frequently, e.g. on daily or hourly basis. Hence, users are required to interact with the EMF User Agent only when there is a change in user's network subscription or there is a change in user's preferred or default network interface. These changes may be affected by a number of parameters e.g. the service levels provided by the network operator, monetary cost of the connections, availability of a particular type of access network, etc. On the basis of these parameters, user can define its default and alternative network interfaces. EMF User Agent must be able to provide the status information of ongoing connections to the user. Status information might include, for example, current data rate, packet loss ratio, estimated time to complete the transfer of current data (e.g. in case of file transfer) and the available alternative links on which the ongoing connections can be migrated. It is the responsibility of EMF Host Agent to provide such information to the EMF User Agent. As this information will be available, the user, if not satisfied by the performance of currently serving link, can initiate the transfer of connection to an alternative link. This is a scenario of willful handover. EMF Core Protocol is responsible to handle connection migration under such willful handovers. To avoid frequent user interventions and to automate the migration procedure, some service level thresholds can be defined. User would be able to change these thresholds any time. Whenever these thresholds are crossed, a notification may be sent to the EMF Host Agent and in turn EMF Host Agent can trigger the EMF association change procedure. D. Interaction between EMF Host Agent & IEEE 802.21 MIH layer IEEE 802.21 is being developed in an effort to standardize an abstraction layer that provides Media

Independent Handover (MIH) function for enabling handover and interoperability between heterogeneous networks [26]. The objective of the standard is to simplify handover management for multihomed mobile devices having not only wired and wireless IEEE network interfaces but also having 3GPP and 3GPP2 cellular network interfaces. MIH function basically provides three services i.e. i) Media Independent Event Service, ii) Media Independent Command Service & iii) Media Independent Information Service. Applications subscribe to the event service, and in response to this subscription, both local and remote events are notified to the upper layer. These events may include "Link parameter change", "Link Up", "Link Down", etc. Command service is responsible to gather the status information of connected links. Commands can be used to poll the connected links for their status and to configure new links. Information service can be used to provide the information about available networks and the services they provide. As shown in Figure 1, EMF Host Agent interacts with 802.21 MIH function for these three services and on the basis of information received by the MIH function, EMF Host Agent can notify the EMF Control module for initiation of EMF association change procedure.

V. HANDOVER MANAGEMENT WITHEMF ARCHITECTURE Whenever EMF Host Agent notices some change i.e. change in IP address or change in user preferences, it signals this change to the EMF Control module. The EMF Control checks the ongoing EMF associations and sends an EMF Association Change message to the peer node. In order to avoid session hijacking attacks, t h ~ s association change message includes the previous association ID. After necessary authentication, peer EMF Control updates its EMF association parameters. A new TCP connection is established with new parameters. EMF Data Handler will maintain sequence numbers for assuring the in-order delivery of user data over multiple TCP connections.

A. Managing Forced Handovers Forced handover scenarios can occur when mobile node is moving out of the serving network and entering in the coverage area of another network. In such situations, user has only one working IP address and has no other option except to migrate the connection to the new network. When mobile node discovers new network and gets new IP address, the EMF Host Agent detects this change and triggers the EMF association change message. B. Managing Willful Handovers Willful handover scenarios can occur when a node has access to more than one network, e.g. simultaneous access to Ethernet, WLAN, WiMax, GPRS, etc. In modern mobile terminals (laptops, PDAs, etc.) it is common to have more than one network interface. A user may wish to use one network interface at one time and other network interface at other time or may wish to use all available network interfaces simultaneously. As discussed earlier, such decisions may be triggered from upper layers. Whenever user manually gives

commands to EMF User Agent to initiate the handover or when serving network's parameters cross the preferences threshold, this information is passed to the EMF Host Agent. EMF Host Agent then notifies the EMF Control to initiate the EMF association change procedure. We assume that the acquisition of new IP addresses for each network interface is implicit. Dynamic Host Configuration Protocol (DHCP) may be used for acquiring IPv4 addresses [32] and in case of IPv6 as underlying network protocol, either DHCPv6 (for tighter control over address assignment) or IPv6 address autoconfiguration can be used for generating globally unique IPv6 addresses [33], [34]. Under already established EMF association, EMF Control initiates new TCP connection using new IP address but with previous association ID.

VI. LOCATION MANAGEMEMENT WITHEMF ARCHITECTURE Handover management, described in previous section, is only one component of the mobility management. The second major component of mobility management is location management of mobile nodes. Although location management may not be the issue for mobile nodes that act only as clients, however, it is a concern for the nodes that also act as server. When a mobile server changes its location, its IP address also changes. DNS dynamic updates provide the facility to dynamically update the name-address mapping of the hosts [35]. BINDv8 and BINDv9 (Berkeley Internet Name Domain) provide the implementation of DNS dynamic update standard [36]. Many mobility management solutions have used the DNS dynamic update option for tracking the location of mobile servers [14], [37]. In most cases, DHCP server sends DNS dynamic update message describing the changes in the assignment of IP addresses. But if there is no DHCP server and IP addresses are generated using IPv6 address autoconfiguration [34], mobile nodes should be allowed to send the DNS dynamic updates. In t h ~ scase, when EMF Host Agent detects the change of IP address, it issues a command to EMF Control to send a DNS dynamic update message. Sending of this DNS dynamic update message doesn't affect the ongoing connection(s) because the problem for maintaining ongoing connection(s) is resolved with EMF handover mechanisms as described in section 4. Location update mechanism is helpful for new connection requests only. In contrast to the initial concerns regarding the scalability and overhead involved in DNS dynamic updates, studes proved viability of using DNS dynamic updates [37]. Incremental zone transfer of DNS data helps to reduce the update overhead [38]. Moreover, Notify option is useful for reducing the delays involved in the dynamic updates [39]. The security concerns regardng DNS dynamic updates have been addresses in [40]. For authentication purposes, secure DNS dynamic updates require the use of Transaction Signature (TSIG) Message Authentication Code (MAC) and SIG(0) public-key encryption. A minor issue regarding the use of shared secret key based TSIG MAC algorithm is that the real time distribution of shared secret key is expensive. So we suggest the use of digital certificates for DNS message

integrity. However, t h ~ s performance optimization of authenticationlintegrity of DNS update messages is out of scope of this article. VII. CONCLUSION In this article, we proposed an end-to-end framework for mobility management for mobile devices. This framework allows the mobile devices to continue their service while moving across heterogeneous wireless networks. The advantage of this framework is that it does not require any supporting intermediate node in the network infrastructure and also it does not require any change in the current implementation of TCP. Another uniqueness of this framework is that the cross-layer design of EMF Host Agent enables the framework to accommodate both forced and willful handovers. EMF Host Agent is able to utilize IEEE 802.21 MIH services. For location tracking of mobile devices EMF Control uses the secure DNS dynamic update option.

Wesley Eddy, "At What Layer Does Mobility Belong", IEEE Communication Magazine, (October 2004) Anand Prasad & Neeli Prasad, "802.11 WLANs and IP Networking; Security, QoS and Mobility", (2005), ISBN 1-58053-789-8, Norwood, MA 02062, ARTECH HOUSE, INC. S. Choi, G. H. Hwang, T. Kwon, A. R. Lim & D. H. Cho, "Fast Handover Scheme for Real Time Downlink Services in IEEE 802.16e BWA System", 61St IEEE Vehicular Technology Conference VTC-2005, Volume 3, Pages 2028-2032, (June 2005) R. Steele, C. C. Lee & P. Gould, "GSM, cdmaOne and 3G Systems", (2001), Print ISBN 0-471-49185-3, John Wiley & Sons Ltd. I. F. Akyildiz, J. Xie & S. Mohanty, "A Survey of Mobility Management in Next-Generation All-IP-based Wireless Systems", IEEE Wireless Communications, (August 2004) J. Manner & M. Kojo, "Mobility Related Terminology", RFC3753, IETF, (June 2004) D. Johnson, C. Perkins & J. Arkko, "Mobility Support in IPv~", RFC3775, IETF, (June 2004) R. Koodli, "Fast Handover for Mobile IPv~",RFC4068, IETF, (July 2005) S. Das, A. McAuely, A, Dutta, A. Misra, K. Chakraborty & S. K. Das, "IDMP: An Intradomain Mobility Management Protocol for Next Generation Wireless Networks", IEEE Wireless Communications, (June 2002) A. Misra, S. Das & S. K. Das, "IDMP Based Fast Handoffs and Paging in IP based 4G Mobile Networks", IEEE Communication Magazine, (March 2002) H. Soliman, C. Castelluccia, K. El Malki & L. Bellier, "Hierarchical MIPv6 Mobility Management", RFC4140, IETF, (August 2005) H. Schulzrinne & E. Wedlund, Application-Layer Mobility Using SIP", ACM Mobile Computing & Communications Review, Vol. 4, No. 3, pp 47-57, (July 2000) C. Brendan, S. Traw & J. M. Smith, "Striping within the network subsystem", IEEE Network Magazine, (July-August 1995) A. C. Snoeren & H. Balakrishnan, "An End-to-End Approach to Host Mobility", 6" ACMIIEEE International Conference on Mobile Computing and Networking, MobiComm'OO, (August 2000) D. A. Maltz & P. Bhagwat, "MSOCKS, An Architecture for Transport Layer Mobility", INFOCOM-98 Seventeenth Annual Joint Conference of the IEEE Computer and Communications Societies, (March-April, 1998) P. Bhagwat, D. A. Maltz & A. Segall, "MSOCKS+, An Architecture for transport layer mobility", Elsevier, Computer Networks, Volume 39, Issue 4, Pages 385-403, (July 2002) H. Y. Hsieh & R. Sivakumar, "A Transport Layer Approach for

Achieving Aggregate Bandwidths on Multi-Homed Mobile Hosts", Springer Journal on Wireless Networks, Issue Volume 11, Numbers 1-2, (January, 2005) R. Stewart et al, "Stream Control Transmission Protocol", RFC2960, IETF, (October 2000) I. Aydin, W. Seok & C. C. Shen, "Cellular SCTP: A Transport Layer Approach to Internet Mobility", Proceedings of 12" International Conference on Computer Communications and Networks, (2003), ICCCN 2003 T. Goff & D. Phatak, "Unified transport layer support for data striping and host mobility", IEEE Journal on Selected Areas in Communications, Volume 22, Issue 4, Pages 737-746, (May 2004) B. Landfeldt, T. Larsson, Y. Ismailov & A. Seneviratne, "SLM, A Framework for Session Layer Mobility Management", In Proceedings of IEEE ICCCN Conference, (1999) J. Rosenberg et al, "Session Initiation Protocol", RFC3261, IETF, (June 2002) N. Banergee, W. Wu, K. Basu & S. K. Das, "Analysis of SIP based mobility management in 4G", Elsevier, Computer Communication, Vol. 27, pp. 697-707, (2004) S. M. Faccin, P. Lalwaney & B. Patel, "IMS, Mobile IP & SIP interaction in 3G", IEEE Communication Magazine, (January 2004) G. Carneiro, J. Ruela & M. Ricardo, "Cross-Layer Design in 4G Wireless Terminals", IEEE Wireless Communication Magazine, (April 2004) F. Cacace & L. Vollero, "Managing Mobility and Adaptation in Upcoming 802.21 enabled Devices", In Proceedings of the 4" International Workshop on Wireless Mobile Applications and Services on WLAN Hotspots, Los Angles, CA, USA, (2006) A. Dutta et al, "Seamless Handover across Heterogeneous Networks - An IEEE 802.21 Centric Approach", In Proc. WPMC 2005, (2005) M. S. Blumenthal & David Clark, "Rethinking the design of the Internet: The End-to-End Arguments vs. the Brave New World", ACM Transactions on Internet Technology, Vol. 1, No. 1 (August 2001) M. Yabusaki, T. Okagawa & K. Imai, "Mobility Management in AllIP Mobile Network: End-to-End Intelligence or Network Intelligence", IEEE Radio Communications Magazine (December 2005) T. Moors, "A Critical Review of End-to-End Arguments in System Design", IEEE International Conference on Communications, ICC2002, Volume 2 (2002) R. Schroeppel, H. Orman, S. O'Malley & 0. Spatscheck, "Fast Key Exchange with Elliptic Curve Systems", In Proceedings of the 15" Annual International Cryptology Conference on Advances in Cryptology, LNCS Volume 963, Pages 43-56, Springer-Verlag, London, UK, (1995) R. Droms, "Dynamic Host Configuration Protocol", RFC 2131, IETF (March 1997) R. Droms, J. Bound, B. Volz, T. Lemon, C. Perkins & M. Corney, "Dynamic Host Configuration Protocol for IPv6 (DHCPV~)",RFC 3315, IETF (July 2003) S. Thomson, T. Narten, T. Jenmei, "IPv6 Stateless Address Autoconfiguration", RFC 4862, IETF (September 2007) P. Vixie, S. Thomson, Y. Rekhter & J. Bound, "Dynamic Updates in the Domain Name System (DNS UPDATE)", RFC 2136, IETF (April 1997) P. Albitz & C. Liu, "DNS and BIND", 4" Edition, Publisher O'Reilly, ISBN 0-596-00158-4 (April 2001) A. Pappas, S. Hailes & R. Giaffreda, "Mobile Host Location Tracking through DNS", London Communication Symposium (September 2002) M. Ohta, "Incremental Zone Transfer in DNS", RFC 1995, IETF (August 1996) P. Vixie, "A mechanism for Prompt Notification of Zone Changes (DNS Notify)", RFC 1996, IETF (January 1996) B. Wellington, "Secure Domain Name System (DNS) Dynamic Update", RFC 3007, IETF (November 2000)