AN EXPOSITION ON WIRELESS IP INTERWORKING

Jon W. Mark, Xuemin Shen and Yu Zeng xshen, yzengg@bbcr.uwaterloo.ca Center for Wireless Communications Department of Electrical and Computer Engineering University of Waterloo Waterloo, ON, Canada N2L 3G1 July 6, 1999
fjwmark,

Abstract
The increasing reliance on information available on Internet, and the rapid growth of the wireless subscriber population suggest a need for Internet users to maintain communications as they move from place to place. However, a mobile user needs to have a stable IP address in order to be stably identi able, and a stable IP address is counter to the concept of mobility. Mobile IP proposed by the Internet Engineering Task Force IETF is the emerging standard for mobile Internet applications. It allows a mobile user to change its location without restarting its applications and without disrupting any ongoing communications. Mobile IP is transparent to the physical medium over which a mobile user communicates. On the other hand, the wireless access to xed wirelined IP networks gives rise to a number of issues. The nature of wireless data links and the user mobility have an evident impact on the performance and usage of IP. This report presents an exposition on wireless IP interworking, a solution to the problem of transferring information through Internet to and from users anywhere and at anytime.

1 Wireless IP Interworking
In recent years, society has witnessed the increasing importance of computers and an increased dependence on them for day-to-day activities. At the same time, communications technology continues to grow rapidly, making possible entities such as the Internet where information from around the world is available to anyone, anywhere at anytime. The most important application of Internet is the World Wide Web which merges computers and communications, and transforms every personal computer into a personal communications device. Other applications of Internet include email, le transfer, e-commerce, customer support, shopping information, work at home, chat, etc.. The proliferation and universal adoption of the Internet as the information transport platform have escalated it as the key wirelined network for supporting xed hosts. The integration of Internet and wireless communication networks is expected to provide multimedia services for both mobile and xed hosts in the near future.
Fixed Host Mobile Host Fixed Host

IP-Based Network
Mobile Host

Wireless Network

Fixed Host

Wireless Network

Figure 1: Wireless IP interworking Figure 1 shows a diagram of Wireless IP interworking. The Wireless IP connections consist of 1 paths or routes through the broadband backbone Internet; 2 radio links between the mobile terminals and base stations or access points in the wireless segment, which provide the interface for mobile users to the xed backbone Internet. Connection establishment and continuous communications between mobile users across the Internet require unique identi cation of Internet Protocol IP addresses. However, in an environment where the mobile end users constantly change their access point, an interworking infrastructure and a networking protocol are needed to support mobility without disrupting any onging communication. Unfortunately, the current suite of internet protocols TCP IP fall short of mobility support since they are designed under the assumption that the end hosts are xed. Currently, there are several proposals to enhance the existing IP to accommodate mobility Bhagwat, et al., 1996; Hills and Johnson, 1996; Perkins, 1997; 2

Solomon, 1998. The Mobile IP proposal from the Internet Engineering Task Force IETF has been evolving towards standardization for Mobile IP. In section 2, we discuss the development of Mobile IP including its standard documentation, functionality, and outstanding issues. The mismatch in transmission speed between the wireless and wired links of the wireless IP networks introduce a networking paradigm with new challenges. The speed mismatch may result in bu er over ow at the attachment points between the Internet and the wireless network. This is compounded by the frequent hando s induced by user mobility. The network control functions of both the Internet and wireless network need to be invoked. In section 3, we examine the impact of wireless link quality and how to improve network control functions. Section 4 provides the authors thought of future research in wireless IP interworking, and conclusions are given in section 5.

2 Mobility Issues
In a wireless environment, the frequent migration of users between radio cells means that the underlying process is nonstationary. As an information transport mechanism, IP resides in the network layer of the protocol stack. IP provides best e ort" service. If a host moves from one location to another without changing its IP address, it will be unable to receive information at the new site. Also, if a host changes its IP address when it moves, it will have to terminate and restart any ongoing sessions. Mobile IP proposed by IETF is designed to o er seamless roaming to mobile hosts who wish to connect to the Internet and maintain communications as they move from one location to another. Mobile IP can be considered to be a routing protocol that allows IP packets to be routed to mobile hosts which may be connected to any link while using their permanent IP address. This is accomplished by allowing mobile users to have two IP addresses, a xed home address for identi cation and a care-of address that changes at each access point for routing. Mobile IP also allows a mobile user to change its location without restarting its connection or disrupting any ongoing communication. In its current form Mobile IP is a network-layer solution to user mobility in the Internet, and is transparent to the physical layer functionalities. Mobile IP has the ability to scale to global networks and to utilize hierarchical distribution of routing addresses to e ectively reduce routing tables. Since Mobile IP can accommodate wireless mobility between networks and simplify mobility across various types of media, it can support heterogeneous mobility. There are data-link layer mobility proposals, such as Cellular Digital Packet Data CDPD and wireless LAN 3

Hills and Johnson, 1996, which also support mobility. CDPD, a standard for delivering IP packets over the North American analog Advanced Mobile Phone Service AMPS cellular telephone network, utilizes idle channels in the cellular system to provide a connectionless digital data packet service. CDPD operates at the physical and data link layers. CDPD is only available in those geographic areas where there is analog cellular coverage by a provider that also support CDPD service, making it more of a some-area" than a wide-area" mobility solution. CDPD has a maximum connection speed around 11 kilobits per second. The North American CDMA and TDMA standards were well underway when CDPD was rst speci ed in the early 1990s. These digital cellular standards will adopt CDPD to provide data service rather than replace it. Compared with CDPD, IEEE802.11 Crow, et al., 1997, a standard for wireless local area networks proposed by the Institute of Electrical and Electronics Engineering IEEE, provides a faster connection speed but within a more limited area. IEEE802.11 supports speeds up to 2 megabits per second. The data-link layer mobility solutions depend on the media for which they are speci ed and require a means for mobility between di erent networks. Therefore, the data-link layer solutions can only provide homogeneous mobility.
2.1 Mobile IP Functionality

Mobile IP introduces the following new functional entities Perkins, 1998: 1. Mobile node is a host or a router which can travel around Internet while maintaining any ongoing communications. A mobile node has a home address, which is a long-term IP address on its home network. When away from its home network, the mobile node is assigned a care-of address which re ects the mobile node's current point of attachment. 2. Correspondent node is a peer with which a mobile node is communicating. 3. Home address is an IP address that is assigned for an extended period of time to a mobile node. It remains unchanged regardless of where the node resides in the wireless segment. 4. A care-of address is the termination point of tunneling datagrams destined to a mobile node while it is away from home. 5. A collocated care-of address is an externally obtained local IP address temporarily assigned to an interface of the mobile node. 4

6. A foreign agent care-of address is an IP address of a foreign agent which has an interface on the foreign link being visited by a mobile node. A foreign agent care-of address can be shared by many mobile nodes simultaneously. 7. Home agent is a router with an interface on the mobile node's home network link which the mobile node keeps informed of its current location, care-of address, as the mobile node moves from link to link. Home agent can intercept packets destined to the mobile node's home address and tunnels them to the mobile node's current location. 8. Foreign Agent is a router with an interface on a mobile node's visited cell which assists the mobile node in informing its home agent of its current care-of address. 9. Link-layer address is an address that identi es the physical endpoint of a link. Usually, the link-layer address is the interface's Media Access Control MAC address. 10. Home network is a network having a network pre x matching that of a mobile node's home address. 11. Foreign network is a network other than a mobile node's home network to which the mobile node is currently connected. 12. Virtual network is a network with no physical instantiation beyond it's router. The router usually uses a conventional routing protocol to advertise reachability to the virtual network. 13. Link is a facility or medium over which nodes can communicate at the link layer. 14. A mobile node's home link is the link which has been assigned the same network-pre x as the networkpre x of the mobile node's home address. 15. A mobile node's foreign link is the mobile node visited link which has been assigned the same networkpre x as the network-pre x of the mobile node's care-of address. 16. Agent advertisement is that foreign agents advertise their presence by using a special message. 17. Agent solicitation is the message sent by mobile nodes to request agent advertisement. 18. A tunnel is the path followed by a datagram while it is encapsulated. 19. Binding entry is an entry in the home agent's routing table. Mobile IP maps the mobile node's home address into its current care-of address. 5

The operation of the Mobile IP is based on the cooperation of three major subsystems: agent discovery, registration, and tunneling routing. 1. Agent Discovery is a process by which a mobile node determines its new attachment point or IP address as it moves from place to place within the wireless segment of the wireless IP network. By agent discovery, a mobile node can i determine whether it is connected to its home link or foreign link; ii detect whether it has changed its point of attachment; and iii obtain a care-of address if it is connected to a foreign link. Mobile node identi es whether it is connected to the home or foreign link from agent advertisements sent periodically by agents home, foreign or both as multicasts or broadcasts to the link. In case that a mobile node does not receive any agent advertisement or it does not have the patience to wait for the next agent advertisement, the mobile node will send an agent solicitation to request an agent advertisement from the agent it is attached. When a mobile node is connected to its home link, it works exactly as a traditional node in a xed place. It routes packets using traditional IP routing protocols. When a mobile node detects that it has moved, it acquires a care-of address by reading it directly from agent advertisement or contacting Dynamic Host Con guration ProtocolDHCP, or using the manual con guration. Registration follows once the mobile node gets a new care-of address. 2. Registration is a process performed as a mobile node enters and remains on a foreign link. This process involves requesting services from a foreign agent and informing the home agent of a mobile node's new care-of address. Registration also involves reregistration upon expiry of a current registration and the deregistration as the mobile node returns to its home link. Some characteristics of registration include having multiple, simultaneous care-of addresses and the ability to remove any number of them while retaining others. Registration consists of an exchange of two messages, a registration request and registration reply between the mobile node and its home agent, possibly a foreign agent be involved as well depending on the type of the mobile node's care-of address. While agent discovery message is carried by the Internet Control Message Protocol ICMP payload portion, registration message is carried by the User Datagram Protocol UDP. Three registration scenarios are: a Mobile node registers using a foreign agent's care-of address which involves a foreign agent. Figure 2 b Mobile node register using a collocated care-of address. Figure 3 6

Registration Request

Mobile Host

Internet
Foreign Agent Mobile Node
Registration Reply

Home Agent

Figure 2: Registration with foreign agent care-of address

Registration Request

Mobile Host

Internet
Home Agent Mobile Node
Registration Reply

Figure 3: Registration with collocated care-of address c Deregistration when a mobile node returns to its home link. Figure 4
Deregistration Request

Mobile Host

Internet
Home Agent
Deregistration Reply

Mobile Node

Figure 4: Deregistration Registration can also serve as a means for a new mobile node to obtain the address of a home agent as it initially con gures itself for Mobile IP. Registration in Mobile IP must be made secure so that fraudulent registrations can be detected and rejected. Otherwise, any malicious user in the Internet could disrupt communications between the home agent and the mobile node by the simple expedient of supplying a registration request containing a bogus care-of address. 3. Tunneling is a process by which Mobile IP tunnels datagrams to and from the mobile node whether it is away or not from its home network. 7

Mobile Host

Mobile Host

Foreign Agent Mobile node with foreigh agent care-of address Mobile Node with collocated care-of address

Internet

Correspondent Home Agent

Packets sent by correspondent Tunnel Packets sent by mobile node

Figure 5: Mobile IP Scenarios Mobile IP operates in the following way Figure 5: Home and foreign agents make themselves known by sending agent advertisement messages. An impatient mobile node may optionally solicit an agent advertisement message. After receiving an agent advertisement, the mobile node determines whether it is on its home network or a foreign network. A mobile node basically works like any other node on its home network when it is at home. When a mobile node moves away from its home network, it obtains a care-of address on the foreign network by soliciting or listening for agent advertisements. The mobile node registers each new care-of address with its home agent, possibly by way of a foreign agent. Datagrams sent to the mobile node's home address are intercepted by its home agent, tunneled by its home agent to the care-of address, received at the tunnel endpoint at either a foreign agent or the mobile node itself, and nally delivered to the mobile node. In the reverse direction, datagrams sent by the mobile node are generally delivered to their destination using standard IP routing mechanisms. There are two type of routing currently proposed: triangle routing and optimized routing. 1. Triangle Routing: Packets that are sent by a correspondent to a mobile node connected to a foreign link are routed rst to the mobile node's home agent and then tunneled to the mobile node's care-of address. On the other hand, packets sent by the mobile node are routed directly to the correspondent. The Mobile IP protocol with triangle routing is simple, the exchange of control messages is limited, and the address bindings are highly consistent since they are kept in one single point for a given host. 8

Mobile Host

Internet
Home Agent
Correspondent

Foreigh Agent

Mobile Host

Internet
Home Agent
Update Process

Foreigh Agent

Correspondent

Packets sent by correspondent Tunnel Packets sent by mobile node Care-of address update

Figure 6: Mobile IP Triangle Routing vs. Optimized Routing One of the drawbacks of the protocol is that the destination home agent is a xed redirection point for exchanging every IP packet, even if a shorter routing path is available between source and destination. This can lead to unnecessarily large end-to-end packet delay. The other drawback is that the network links connecting a home agent to the network can easily be overloaded. Indeed, all session paths sharing the subnet eld of their destination address converge into that subnet home agent, even if adjacent network links are idle. 2. Optimized Routing: The mobile node informs the correspondent node of its care-of address and have the packets tunneled directly to the mobile node, bypassing the home agent. The Mobile IP protocol with optimized routing allows every tra c source to cache and use binding copies. It supports a further update process by which a binding copy can be propagated to the requiring nodes, which may keep it in their cache for immediate or future use. Local bindings enable most packets in a tra c session to be delivered by direct routing, with apparent gain in terms of quality of service and scalability. In addition, a moving host can always inform its previous foreign agent about the new care-of address, so that packets tunneled to the old location owing to an out-of-date binding copy can be forwarded to the current location. This should increase overall quality of service in case of high mobility. The disadvantage of the protocol is that it is quite complex, exchange of control messages and processing overhead due to cache queries could be critical, and cached bindings are possibly inconsistent since they are kept in a distribution fashion. The main obstacle to implement optimized routing resides in 9

the security issues. The correspondent node must be informed of the mobile node's care-of address in order to tunnel data to the mobile node. In an hostile environment, an intruder can easily cut o all the communication to the mobile node by sending a bogus registration if he knows the mobile's care-of address. Therefore, authentication has to be introduced in optimized routing. Since Mobile IP requires the transmission of routing updates between the various nodes in the network, it is important to make the size and the frequency of these updates as small as possible in order to make the Mobile IP suitable for use over wireless links. The Mobile IP mechanism should also be su ciently simple and exible to accommodate mobile node softwares. This increases the number of mobile nodes which can potentially make use of Mobile IP pagers, smart cellular phones and personal organizers, etc.. It is important to make Mobile IP to avoid solutions which require mobile nodes to use multiple addresses since Internet is running out of available IP addresses. In summary, Mobile IP is a peripheral mechanism to and has minimal impact on the Internet since it de nes only three entities where the protocol needs to be implemented: the mobile node, the home agent, and the foreign agent. Mobile IP has the ability to allow the mobile node to change its point-of-attachment from one link to another while maintaining all existing communications. Mobile IP also provides communication between an internet user and a mobile node transparent to the movement of the mobile user.
2.2 Outstanding Mobile IP Issues

The most outstanding problems facing Mobile IP are related to security and rewalls. 1. Security refers to the protection of computers, network resources, and information against unauthorized access, modi cation, and or destruction. This generally involves four related topics: con dentiality transforming data such that it can be decoded only by authorized parties; authentication - proving or disproving someone's or something's claimed identity; integrity checking - ensuring that data cannot be modi ed without such modi cation being detectable; and non-repudiation - proving that a source of some data did in fact send data that he might later deny sending Solomon, 1998. The technology employed to accomplish all of these security features is called cryptography. Cryptography is the science of transforming data in seemingly bizarre ways to accomplish surprisingly useful things. A cryptographic system consists of two fundamental components: a complicated mathematical function 10

- algorithm, and one or more secret or public values - keys. A key is a group of binary data that is known only to the parties which wish to communicate securely. An algorithm is usually published and is available to anyone who wants to read it. There are numerous security protocols and cryptographic algorithms in use throughout portions of the global Internet Solomon, 1998. The protocols that provide solutions to the security problems introduced by mobility can be found in RFCs 1825 - 1827 Atkinson, 1995. Because registration packets are sent remotely by the mobile node to change the home agent's routing table, the home agent must be certain that the registration is indeed sent by the rightful mobile host. Moreover, mobile IP generally works in wireless networks which are more vulnerable to passive eavesdropping, active reply attacks and destruction. Authentication registration is accomplished by using Keyed Message Digest 5 KMD5. Each pair of mobile node and home agent must share a security association and be capable of creating unforgettable digital signature for registration request using 128 MD5 keys. Mobile IP includes within the registration message an identi cation eld that changes with every new registration to prevent a malicious node from recording a valid registration for later use. Two ways, timestamping and pseudo-random numbering, are generally used to ensure the uniqueness of the identi cation eld. 2. A rewall is a device or a set of devices which separates a trusted, private network from an untrusted, public network. A rewall protects a private network from intrusion by outsiders on the public network, but it does not prevent insiders from exchanging information with others on the public network. Three primary types of rewalls are packet ltering routers, application-layer relays, and secured tunnelers. Using a home address as the source address in packets could raise suspicion in rewalls resulting in discarded packets. Typically, packets with an address that makes them look like they should arrive from a certain direction but actually arrive from another, are discarded. Also, a mobile host transmitting within a secured domain may not be able to send packets outside if it has an unexpected source address. Firewalls complicate an Internet-wide deployment of Mobile IP, since they are speci cally designed to prevent unauthorized access to private networks from external portions of the Internet. The Mobile IP working group is in the process of de ning procedures by which mobile nodes can send and receive packets through the rewalls while protecting their private networks without compromising the security of those networks. These procedures use the IP security protocols - the Authentication Header and Encapsulating Security Payload - to realize authenticated and encrypted tunnels between a mobile node and its rewall Solomon, 1998.

11

2.3

IP Version 6

As the development of global internet evolves, the existing Internet Protocol version 4 IPv4 becomes obsolete and creates a number of problems for administration and operation. The available IP addresses will soon be dried up because of the rapid growth of hosts connected to the internet and the current address allocation scheme. The requirements for network routers grow rapidly in terms of memory and performance. It is necessary to design the new generation Internet Protocol which can not only transfer the current IPv4 to the new protocol but also solve the issues such as addressing, routing, security, mobility, Quality of Service QoS, etc. Since 1994, the Internet Engineering Task Force IETF has been working on the next generation internet protocol, IPv6 Stallings, 1996. Some o cial IPv6 documents are listed in Appendix B. As a successor to the IPv4 the basic IPv4 header is shown in Figure 7, IPv6 can be installed as a normal software upgrade in Internet devices and is interoperable with the current IPv4. The IPv6 can be run well on high-performance networks ATM, fast Ethernet, etc. and is at the same time, e cient for low-bandwidth networks. Stateless address autocon guration and neighbor discovery are part of IPv6's support for mobility. The basic IPv6 header is shown in Figure 8.
Version 4 bits Header Length Type of Service 8 bits Flag 3 bits Total Length of Datagram 16 bits Fragment Offset (13 bits) Header Checksum 16 bits Datagram Identification (16 bits) Time to Live 8 bits Protocol 8 bits

Source IP Address (32 bits) Destination IP Address (32 bits) IP Options

Data Portion of Datagram

Figure 7: IPv4 Header The changes from current IPv4 to IPv6 fall primarily into the following categories Hinden, 1996: 1. Expanded Routing and Addressing Capabilities: With a 32- bit IP address in IPv4, which can hold up to 232 , 1, over 4 billion hosts, one might think this address range is more than enough to support the address need on Internet. However, we can easily run out of available addresses because: 12

Version 4 bits

Priority 4 bits Payload Length (16 bits)

Flow Label (24 bits) Next Header (8 bits) Hop Limit (8 bits)

Source IP Address (128 bits)

Destination IP Address (128 bits)

Figure 8: IPv6 Header a Although the traditional two-level address scheme, network pre x and host address, is convenient, there is a waste of address space. Once a network address is assigned to a particular network, a block of IP address is assigned to that network. Any improper assignment of network address will lead to a great waste of available IP addresses, especially for class A and class B network. b Many private networks, which are currently not connected to the internet, are reusing IP addresses used by public network or other private networks. They require much more IP addresses when connected to internet c There may be other kinds of devices other than traditional host, possible wireless and wirelined products, such as mobile telephone, wireless organizer, etc., need additional IP address to make themselves identi ed on the internet. The increase of IP address size from 32 bits to 128 bits in IPv6 allows it to support more levels of addressing hierarchy, a much greater number of addressable nodes, and simpler autocon guration of addresses. The traditional two-level IP address structure, network address and host address, is modi ed. Unicast address, anycast address and multicast address are de ned in the IPv6. a Unicast address: simply a 128-bit network node address. Unicast address can be divided into two parts: a subnet pre x, which indicates the node's subnetwork, and an interface ID, which indicates the node's interface. Unicast address ranges are also allocated for non-TCP IP network such as Novell's IPX and ISO internetwork protocol. b Multicast Address: multicast address in IPv6 will replace broadcast address in IPv4. Multicast 13

address is divided into two groups: the prede ned groups, which are permanently assigned, and the transient groups, which are de ned by speci c organizations. The most common prede ned multicast addresses are All Node - refer to all nodes connected, both routers and hosts, All Routers - which do not include hosts, and All Host - which do not include routers. c Anycast address: a new type of address de ned to identify sets of nodes where a packet sent to an anycast address is delivered to any one of the nodes assigned that address. Packets destined to a multicast address are sent to all nodes in that group while packets destined to an anycast address are sent to only one node in that group. The use of anycast address in IPv6 allows nodes to control the path along which their tra c ows. The scalability of multicast routing is improved by adding a scope" eld to the multicast address. There will be coexistence of both IPv6 addresses and IPv4 addresses and, also, it is impossible to replace all IPv4 routers with IPv6 routers. In IPv6, special addresses, called IPv4-compatible-IPv6 addresses, are introduced to be assigned to those hosts and routers running IPv6 but must route tra c across IPv4 networks. While IPv4-mapped-IPv6 addresses are assigned to those hosts running IPv4. 2. Header Format Simpli cation: IPv6 header has a xed length of 40 octets. Some IPv4 header elds have been dropped or made optional, to reduce the common-case processing cost of packet handling and to keep the bandwidth cost of the IPv6 header as low as possible despite the increased size of the addresses. IPv6 extension headers, optional parts following the IPv6 header, are de ned to carry additional information about the tra c being sent. Extension headers include a Hop-by-Hop option header - de nes special options that require hop-by-hop processing b Fragment header - Contains fragmentation and reassembly information c Destination options header - Contains optional information to be examined by the destination node d Routing header - provides extended routing e Authentication header - provides packet integrity and authentication 3. Improved Support for Options: Changes in the way IP header options are encoded allows for more e cient forwarding, less stringent limits on the length of options, and greater exibility for introducing new options in the future. 14

4. Priority: A 4-bit priority eld is introduced for the source node to indicate the desired transmit and delivery priority of every packet relative to other packets from the same source. Tra c types are rst classi ed as congestion-controlled tra c or non-congestion-controlled tra c, then one of eight levels of relative priority is assigned to each kind of tra c. Congestion-Controlled-Tra c refers to tra c which can tolerate congestion, or delay. If network congestion happens, congestion-controlled-tra c will be bu ered or back o ". A variable amount of packet delay or even out-of-order packet arrival is acceptable. IPv6 de nes the following types of congestion-controlled tra c, in decreasing priority: internet control tra c, interactive tra c, attended bulk transfer, unattended data transfer, ller tra c and uncharacterized tra c. Non-Congestion-Controlled-Tra c refers to those tra c which require constant data rate, constant delivery rate, or at least relatively smooth data rate or delivery delay. Examples of non-congestioncontrolled tra c are real-time audio and real-time video. 5. Quality-of-Service Capabilities: A new capability is added to enable the labeling of packets belonging to particular tra c ows" for which the sender requests special handling, such as voice or video. A ow is basically a series of packets originated by the source and have the same transmission requirements. No special transmission process is assigned to any particular ow label. A source must specify or negotiate what kind of special handling requested before a ow is transmitted, possibly by means of other internet control protocol. Flow labels are assigned pseudo-randomly to ensure that there is no ow label reuse during the life-time of that ow label. 6. Security Capabilities: IPv6 will support the ve proposed security-related standards published by IETF. These security features, which are optional in IPv4, are Atkinson, 1995: RFC 1825 - Security Architecture for the Internet Protocol RFC 1826 - IP Authentication Header RFC 1827 - IP Encapsulating Security Payload ESP RFC 1828 - IP Authentication Using Keyed MD5 RFC 1829 - The ESP DES-CBC Transform Two IP security mechanisms, security association and authentication, are combined to transmit IP packets that require both privacy and authentication. IPv6 includes the de nition of extensions that provide support for authentication, data integrity, and con dentiality. This is included as a basic element of IPv6 and will be included in all implementations. 15

Although a Mobile IPv6 node can easily acquire an IP address and begins communication, it does not change nature of network-pre x routing. Therefore, IPv6 need a mobility solution for exactly the same reason as IPv4. Mobile IPv6 concepts are similar to those of Mobile IPv4, including mobile nodes, home agent, home address, and care-of address, but are more re ned and fully integrated with stationary functions. The current activities by the IETF on Mobile IP are contained in IPv6. Mobile IPv6 uses the new-and-improved IPv6 Routing Header, along with the Authentication Header and other pieces of IPv6 functionality, to simplify routing to mobile node and to perform route optimization in a secure fashion. Mobile IPv6 has no foreign agent. The mobile node uses the Address Autoconfiguration procedure de ned in IPv6 to acquire a collocated care-of address on a foreign link, and reports its care-of address to its home agent and selected correspondents. The mobile node noti es its home agent when it returns to its home link. In Mobile IPv6, correspondent nodes which know a mobile node's current care-of address can send packets directly to the mobile node by using an IPv6 Routing Header. Correspondent nodes not possessing this information send packets without such a header where they are routed to the mobile node's home link, intercepted by the home agent, and tunneled to the mobile node's care-of address  gure 9.
Mobile Host

Correspondent which knows the care-of address

Mobile node

Internet

Home Agent

Correspondent which doesnt know the care-of address Packets sent by correspondent Tunnel Packets sent by mobile node Source Routing

Figure 9: Mobile IPv6 Scenarios Since IPv6 standards documents are still at an early stage of standardization, more requirement for mobility enhancement may be placed on IPv6 nodes. With Mobile IPv6 application, Internet service o erings include 16

1. Internet Telephony Voice-over-IP 2. Virtual Private Networks 3. Quality-of-Service 4. E-commerce 5. Security The advantages of Mobile IPv6 over the Mobile IPv4 are: 1. The enormous address space in IPv6 allows very simple address autocon guration by means of Stateless Address Con guration SAC - combine the network-pre x with the mobile node's interface token, the link-dependent identi er. Because a mobile node can easily get collocated care-of address by SAC, foreign agent functionality is no longer needed. As a result, foreign agent is gone in mobile IPv6. This also implies that all mobile IPv6 care-of addresses are collocated care-of address. 2. Mobile IPv6 uses new IPv6 routing header to simplify routing to mobile nodes. 3. With the enhanced authentication header in IPv6 and the mandatory implementation of IP authentication header, mobile IPv6 might adopt a widescale of route optimization if a key management infrastructure becomes widely available on the internet. 4. Mobile IPv6 has almost the same terminologies as mobile IPv4 except that the concept of foreign agent does not exist any more. The concept of home agent, home link, care-of address and foreign link are roughly the same as in mobile IPv4 5. Mobile IPv6 uses both tunneling and source routing to deliver packets to mobile nodes, the former being the only mechanism used in mobile IPv4.

3 Wireless Access
Wireless access to the Internet introduces a di erent set of issues from the conventional wired access. Compared to wired links, wireless links exhibit high bit error rates BERs. Environmental factors and location of a mobile host determines link quality. Since the wireless channel is interference limited, data transmission 17

speeds over the wireless links are normally well below wired media. Mobile IP is a network layer protocol dealing with mobility from a network point of view. Providing wireless access to Internet unveils issues e ected by wireless link characteristics and wireless mobility. These issues, which involve the physical layer of the protocol stack and have adverse e ect on end-to-end performance, are not accounted for by Mobile IP. A salient feature of a wireless environment is its ability to support roaming. To do so an e cient hando mechanism is necessary. Hando ensures the continuity of the connection as the user roams. The provision of hando procedures can have great impact on system capacity and network performance. The steps involved in the construction of a hando algorithm may include measurement, hando initiation and execution, and resource allocation. Resource management to minimize hando overhead within the wireless domain is already a complex problem. The inclusion of the internet for interconnection further increases this complexity. The frequent migration of users among base stations renders the underlying process nonstationary. A meaningful set of processes to describe the mobility of users across the service area is required. This is particularly relevant in view of the intended support of multimedia tra c. Satisfaction of di erent QoS requirements will necessitate e ective and e cient network access control, which may have to adapt to the changing environment induced by user mobility and time-varying channel conditions.

Wireless Link Issues

Wireless links often experience long-term and short-term fadings. Long-term fading occurs as the average or the local mean of the fading signal changes slowly with time. Short-term fading is also known as multipath fading which occurs due to refractions by multiple scatterers. Both types of fading lead to signal degradation and an increased bit error rate BER, and hence an increased packet-error-rate Jakes, 1994. Internet is typically designed with a packet loss probability less than 10,6. The quality of a wireless link, with a typical BER of 10,2 to 10,4, is poor. Forward error correction FEC and antenna diversity schemes used to cope with the bit- and packet-error-rate problem at the physical layer.1 These two techniques add complexity and cost to the mobile end system since FEC requires encoding decoding circuitry, and antenna diversity requires digital signal processing. A combination of the two techniques may be used to tailor a system's cost performance tradeo for a given radio environment Naghshineh et al., 1996. The di erence in transmission rate between wired and wireless links is sometimes great enough to cause
1

FEC decoding is normally performed at the receiver. Even detection is performed at the data link layer.

18

congestion at the attachment points. Congestion control schemes to regulate the tra c ow from high speed links to low speed links are reeked.2 The role of such schemes is to control the transmission of packets from the high-speed network in order to reduce the bu er over ow at switches of low speed networks. Wireless communication links are scarce resources because of the limited bandwidth. To make use of the most of the wireless links, it is necessary to develop services using a Mobile Application Framework MAF". MAF should support conventional applications a by optimizing outgoing communications call and providing means for transparent use of di erent protocols; b supporting disconnected or o line operation on personal data les through a le logging system which logs changes to data les, and periodically reconciles these changes into other replicas held at other locations; c allowing the user to specify a monetary communications budget, and by scheduling reconnection events such that the budget is closely met. A new transparent communication layer, called TACO, has been introduced. It schedules packet transfer at the system level and at the application level. The system level functions arbitrate the resource among competing applications, whereas application level scheduling allows applications to adapt to changes in resources. TACO decides which packets from which application to send, by using priority and latency related interface pro les.

Modi ed TCP
TCP Transport Control Protocol is a connection-oriented transport layer protocol that is designed to provide reliable and in-order data delivery. However, if TCP is used without any modi cation in wireless networks, a serious drop of throughput may be expected. The reason is that high bit errors or even disconnections due to the poor quality of wireless links can corrupt packets, which may result in losing TCP data segments or acknowledgments. When acknowledgments do not arrive at the TCP sender within a prescribed interval of time, a timeout occurs. The sender retransmits the segment, exponentially backs o its retransmit timer for the next retransmission, and then reduces its window to one segment. Repeated errors will cause the window to remain small resulting in low throughput, especially on long links. It is important to note that FEC may be used to combat high BER, but it will consume scarce wireless bandwidth when correction is not necessary. In order to ensure that the TCP connection to a mobile is e cient, it is necessary to prevent the sender from shrinking its congestion window when packets are lost either due to BER or due to disconnection. When the mobile is reconnected, it should begin to receive data immediately. Several proposals have been made for a
2

Tra c ow from high-speed links to low-speed links can cause congestion and home bu er over ow, but not vice versa.

19

new TCP protocol that is optimized for use over wireless links Balakrishnan, et al., 1996. These include Indirect Transmission Control Protocol I-TCP, Berkeley Snoop Module, Fast Retransmit, TCP for Mobile Cellular Network M-TCP. In the I-TCP protocol Bakre and Badrinath, 1996, any connection between a mobile host and a xed host is split into two separate connections at the base station - one between the mobile host and the base station or its mobile support router MSR over the wireless medium, and the other between the base station or MSR and the xed host over the wired network Figure 10. In this way, the special requirements of mobile hosts can be accommodated that is backward compatible with the existing xed network. All the specialized support that is needed for mobile applications and for the low speed and unreliable wireless links can be built into the wireless side of the interaction while the xed side is left unchanged. Data sent to the mobile host is received and acknowledged by the base station before being delivered to the mobile host. The wireless wired link characteristics would be hidden from the transport layer, and only the wireless resource would be used for error control when the error is caused by the wireless link. With the I-TCP protocol, the resulting bene ts are: a it separates the ow control and congestion control functionality on the wireless link from that on the wired link; b a separate transport protocol for the wireless link can support noti cation of events such as disconnections, moves and other features of the wireless link e.g., available bandwidth, to the higher layers; c partitioning the connection into two distinct parts allows the base station to manage much of the communication overhead for a mobile host. Throughputs are increased with the I-TCP since the node, where the connection is split, may be one or two hops away from the mobile host's cell and can adapt more quickly to the dynamic mobile environment because the round trip time is shorter. However, I-TCP does not maintain end-to-end TCP semantics. This is because the TCP acknowledgments are not end-to-end but instead there are two separate acknowledgments for the wireless and wired parts of the connection. One consequence is that the sender may believe a segment is delivered correctly to the mobile host since the base station acknowledged it even if the mobile host is disconnected before receiving this segment. In other words, the sender does not know whether packets are actually received by the mobile host, and this may be a serious problem for many applications. The Berkeley Snoop Module Balakrishnan, et al., 1996 is another proposed solution for losses caused by high BER. The Berkeley Snoop Module makes changes to network layer software at the base station. It caches packets at the base station and inspects the TCP header of TCP data packets and acknowledgments which pass through and bu ers copies of the data packets. Using the information from the headers, the 20

Mobile Host

Internet
Mobile Support Router
Wireless TCP Regular TCP

Correspondent

Figure 10: Indirect Transmission Control Protocol I-TCP snoop module detects lost packets a packet is assumed lost when duplicate acknowledgments are received and performs local retransmissions across the wireless link to alleviate problems caused by high BER. The module also implements its own retransmission timer, similar to the TCP retransmission timeout, and performs selective retransmissions when an acknowledgment is not received within this interval. Routing protocol is also modi ed to enable low-latency hando to occur with negligible data losses. Experiments have shown the Berkeley Snoop Module achieves throughput up to 20 times over and above that of regular TCP and hando latencies over 10 times shorter than that of other mobile routing protocols. The drawbacks of the snoop module is that it does not perform as well either in the presence of lengthy disconnections or in environments where there are frequent hando s. If the mobile host is disconnected for a lengthy period of time, the sender will automatically invoke congestion control because it will not have received acknowledgments for some segments. The snoop module will persistently generate persist packets and these packets will serve no purpose since the mobile is disconnected. If a mobile host moves into a new cell, the new base station starts up a copy of the snoop module on behalf of this mobile host. This snoop module starts out with an empty cache and slowly builds the cache up, and the mobile host will see poor TCP throughput. If the cell sizes are small, the performance degradation can be serious. Fast Retransmit Caceres and Iftode, 1994 is proposed to combat the e ects of short disconnections on TCP throughput. During a hando , since the mobile host cannot receive packets, unmodi ed TCP at the sender will think a congestion has occurred and will begin congestion control reduce window size and retransmit after a time out. The timeout period may be long, even though the mobile host may have completed the hando it will have to wait for the full timeout period before it can begin receiving packets from the sender. Fast Retransmit forces the mobile host to retransmit, in triplicate, the last old acknowledgment as soon as it nishes a hando . This forces the sender to reduce the congestion window to a half and retransmit one segment immediately. Fast Retransmit does not split the TCP connection. However, if the mobile host 21

were disconnected for a long time, the sender would already have invoked congestion control and shrunk its window to one segment. Similarly, if disconnections are frequent or the wireless links are poor, Fast Retransmit will do little to improve throughput because the sender's congestion window will repeatedly get shrunk to half of its previous size. TCP for Mobile Cellular Network M-TCP Brown and Singh, 1997 works in a three-level hierarchy: mobile hosts, supervisor host, and Internet from the lowest to the highest Figure 11, by introducing Supervisor Host SH. Several base stations are controlled by a SH which is connected to the Internet and handles most routing and other mobile users' requirement. The advantages to use this hierachy are: 1. When a mobile host roams from one cell to another, the two base stations do not need to transfer any state information if they are controlled by the same SH. 2. The roaming mobile host remains within the domain of the same SH for long time periods because several base stations are controlled by the SH. By introducing SH, M-TCP maintains end-to-end TCP semantics while it delivers excellent performance when mobile hosts encounter disconnection. This is done by splitting the TCP connection at the supervising host. As packets arrive from a sender in the Internet, an acknowledgment is sent back and the supervisor host deals with ensuring the completion of delivery. Experimentation has shown that M-TCP delivers excellent performance for environments where the mobile encounters periods of disconnection. The drawback of this scheme is the complexity. Also, shortage of bu er space is likely if a supervisor host services several mobile nodes.
Internet
Level 1

Supervisor Host

Supervisor Host

Level 2

Base Station

Level 3

MH

MH

MH

Figure 11: M-TCP hierarchy 22

Modi ed UDP
User Datagram Protocol UDP is a datagram communication service built on top of IP. It adds multiplexing and error detection to the IP capabilities. In contrast to TCP, UDP does not use acknowledgments; it does not retransmit erroneous packets or control the ow. In the wireless IP interworking, a large percentage of packets will be lost by using UDP over wireless links. This is because UDP will continue to send packets even when transmission to a mobile host experiences signal fading. A simple concept would be to stop sending datagrams to a mobile host once it encounters fading. The goal of creating a new modi ed UDP M-UDP is to ensure that packets that were lost are resent. In the M-UDP protocol Brown and Singh, 1996, the UDP connection is split in two at some host close to the mobile host. The host attempts to use any free bandwidth to retransmit packets lost during a fade, thus ensuring that the number of lost packets is kept small. The mobile network architecture is the same as given in M-TCP. Experimentation has shown that the unmodi ed UDP has a loss rate of up to 50 while M-UDP has a loss rate of less than 5 for a wide range of loads and fade intervals.

4 Future research
Future research on wireless IP interworking should focus on seamless delivery of multimedia tra c to and from mobile users, taking into consideration the e ects of both user mobility and wireless link quality. The main concerns are the network and transport layer functionalities. It is possible to use measurements such as pilot signal power to estimate the channel state information and to use the resultant estimates to adjust control function; for example, maintaining transport layer connectivity at the point of attachment of a mobile user to network. Further, investigation should be made into wireless IP interworking to enhance the capabilities of the cellular system, following current standards such as Global System Mobility GSM or IS-95 as well as the target Code Division Multiple Access CDMA systems. Adaptive or dynamic resource management algorithms should be taken into consideration for future network layer protocol development, in order to improve the e ciency of the limited wireless bandwidth. One approach could be to divide a multimedia connection into multiple substreams, each having its own QoS parameters. When the connection capacity shrinks, streams with lower priority would be suspended and, if in transit, dropped to reduce the tra c load. For example, a video transmission could be separated into multiple streams, the aggregate of which would constitute the highest quality video and a lower quality subset. 23

As network conditions improve, suspended streams would be reactivated. Network elements switches, access points, services, and protocols signaling, control, routing would need to be made aware of the QoS requirements for complete integration Naghshineh and Willebeek-leMair, 1997. Performance evaluation requires information on the radio resource management, user mobility model, call admission control, and ow control strategies at the network and link layers. The relation between multimedia QoS parameters, and network and link layer performance metrics should be de ned and speci ed. At the wireless network and Internet interface, the required capabilities should include veri cation of the service class information of the called user, acquisition of location information and interpretation of the routing information and rerouting of the call, when appropriate. Viruses and hackers attacks are common in current Internet. Security issue is very critical for wireless IP interworking because it is more vulnerable for wireless networks to be attacked by hostile intruders. Designing more robust networking is necessary to support commercial activity on Internet.

5 Conclusions
The purpose of this report is to provide an exposition on the current state of Mobile IP and to explore issues related to wireless access and wireless IP interworking. Mobile IP is a comprehensive solution to handle user mobility for the Internet. Before the deployment of wireless access networking becomes a reality, the issues pertaining to the wireless domain, as identi ed in Section 3, must be overcome. Although some proposals to deal with the aforementioned issues have reported in the literature, investigation into e ective and e cient wireless wirelined interworking is necessary.

Acknowledgment
This work has been supported by a grant from the Canadian Institute for Telecommunications Research CITR under the NCE program of the Government of Canada.

24

Appendix A Standard Documents of Mobile IP

Mobile IP has progressed along the ladder to standardization within the IETF, and its speci cations are available as Request for Comments RFC. RFCs can be obtained on Web page, ftp: ftp.isi.edu innotes rfc.txt, where  is the actual number of the desired RFC. The speci cations of Mobile-IP are: RFC 2002 - IP mobility support Perkins, 1996 RFC 2003 - IP encapsulation within IP Perkins, 1996 RFC 2004 - minimal encapsulation within IP Perkins, 1996 RFC 2005 - applicability statement for IP mobility support Solomon, 1996 RFC 2006 - the de nitions of managed objects for IP mobility support Cong, et al., 1996 RFC 1701 - generic routing encapsulation Hanks, et al., 1994 RFC 1853 - IP in IP tunneling Simpson, 1995 RFC 2344 - Reserve tunneling for mobile IP Montenegro, 1998 RFC 2356 - Sun's SKIP rewall traversal for mobile IP Montenegro and Gupta, 1998 Internet Drafts, which are the preliminary documents produced by the various working groups of the IETF, are available at ftp: ftp.ietf.org internet-drafts draftname.txt, where draftname is the name of the desired internet draft. Internet Drafts are only available for a period of six months after which they are revised, published as RFCs, or otherwise deleted. The current Internet Drafts related to the Mobile IP speci cations are: Route Optimization in Mobile IP Perkins and Johnson, 1999 Mobility Support in IPv6 Johnson and Perkins, 1998 Firewall Support for Mobile IP Montenegro and Gupta, 1998 Registration Keys for Route Optimization Perkins and Johnson, 1997 Special Tunnels for Mobile IP Perkins and Johnson, 1997 25

Tunnel Establishment Protocol Calhoun, Montenegro and Perkins, 1998 Rapid Authentication for Mobile IP Sanchez and Troxel, 1997 Use of IPSec in Mobile IP Zao and Condell, 1997 Support for Mobile IP in Roaming Aboba, 1998 Mobile IP Dynamic Home Address Allocation Extensions Calhoun and Perkins, 1998 Mobile IP Regionalized Tunnel Management Calhoun, Montenegro and Perkins, 1998 Mobile IP Challenge Response Extension Calhoun and Perkins, 1999 Mobile IP Network Address Identi er Extension Calhoun and Perkins, 1999 NAI Resolution for Wireless Networks Aravamudhan, Brien, and Patil Requirements on Mobile IP from a Cellular Perspective Gustafsson, Jonsson, Hubbard, Malmkvist, and Roos IP Mobility Architecture Framework Becker, Patil, and Qaddoura Simple Mobile IP SMIP Johnsson AAA Registration Keys for Mobile IP Perkins and Calhoun Mechanism to Support CHAP Mobile Node Authentication for RADIUS DIAMETER Hybrid AAA Networks Xu, Peirce, and Campbell IP micro-mobility support using HAWAII Ramjee, Porta, Thuel, Varadhan, and Salgarelli Paging support for IP mobility using HAWAII Ramjee, Porta, and Li Private Addresses in Mobile IP Perkins, Montegegro, and Calhoun

Appendix B Standard Documents of IPv6

RFC 1886 - DNS Extensions to support IP version 6 Thomson and Huitema, 1995 RFC 1887 - An Architecture for IPv6 Unicast Address Allocation Rekhter and Li, 1995 RFC 2373 - IP version 6 Addressing Architecture Hinden and Deering, 1998 26

RFC 2374 - An IPv6 Aggregatable Blobal Unicast Address Format Hinden, O'Dell and Deering, 1998 RFC 2375 - IPv6 Multicast Address Assignments Hinden and Deering, 1998 RFC 2452 - IP version 6 Management Information Base for the Transmission Control Protocol Daniele, 1998 RFC 2454 - IPv6 Management Information Base for the User Datagram Protocol Daniele, 1998 RFC 2460 - Internet Protocol, version 6 IPv6 Speci cation Deering and Hinden, 1998 RFC 2461 - Neighbor Discovery for IP version 6 Narten, Nordmark and Simpson, 1998 RFC 2462 - IPv6 Stateless Address Autocon guration Thomson and Narten, 1998 RFC 2463 - Internet Control Message Protocol ICMP6 for IPv6 Speci cation Conta and Deering, 1998 RFC 2466 - Management Information Base for IP version 6 Haskin and Onishi, 1998 RFC 2467 - Transmission of IPv6 Packets over FDDI Networks Crawford RFC 2470 - Transmission of IPv6 Packets over Token Ring Networks Crawford, Narten, and Thomas RFC 2471 - IPv6 Testing Address Allocation Hinden, Fink, and Postel RFC 2472 - IP Version 6 over PPP Haskin, and Allen RFC 2473 - Generic Packet Tunneling in IPv6 Speci cation Conta and Deering RFC 2497 - Transmission of IPv6 Packets over ARCnet Networks Souvatzis RFC 2507 - IP Header Compression Degermark, Nordgren, and Pink RFC 2526 - Reserved IPv6 Subnet Anycast Addresses Johnson and Deering RFC 2529 - Transmission of IPv6 over IPv4 Domains without Explicit Tunnels Carpenter and Jung RFC 2553 - Basic Socket Interface Extensions for IPv6 Gilligan, Thomson, bound and Stevens

27

References
B. Aboba, Support for Mobile IP in Roaming", draft-ietf-roamops-mobileip-01.txt, Mar. 1998. R. Atkinson, Security Architecture for the Internet Protocol", RFC 1825, Aug. 1995. R. Atkinson, IP Authentication Header", RFC 1826, Aug. 1995. R. Atkinson, IP Encapsulating Security Payload ESP", RFC 1827, Aug. 1995. R. Atkinson, IP Authentication", RFC 1827, Aug. 1995. R. Atkinson, The ESP DES-CBC transform", RFC 1827, Aug. 1995. A. Bakre and B. R. Badrinath, I-TCP: Indirect TCP for Mobile Hosts", Proc. 15th Inter. Conf. Distributed Computing Systems, Vancouver, Canada, June 1995. A. Bakre and B. R. Badrinath, Indirect Transport Layer Protocols for Mobile Wireless Environment", Mobile Computing, Kluwer Academic Publishers, 1996. H. Balakrishnan, S. Seshan, and R. Katz, Improving Reliable Transport and Hando Performance in Cellular Wireless Networks", Wireless Nteowrks, Dec. 1995. H. Balakrishnan, V. N. Padmanabhan, S. Seshan, and R. H. Katz, A Comparison of Mechanisms for Improving TCP Performance over Wireless Links", IEEE ACM Transactions on Networking, Vol. 5. pp. 756-769, Dec. 1997. P. Bhagwat. C. Perkins, and S. Tripathi, Network Layer Mobility: An Architecture and Survey", IEEE Personal Commu., June 1996. K. Brown and S. Singh, M-UDP: UDP for Mobile Cellular Networks", Computer Communication Review, Oct. 1996. K. Brown and S. Singh, M-TCP: TCP for Mobile Cellular Networks", Computer Communication Review, Oct. 1997. R. Caceres and L. Iftode, Improving the Performance of Reliable Transport protocols in Mobile Computing 28

Environments", IEEE JSAC, Vol. 13, June 1994. P. R. Calhoun, G. Montenegro, and C. Perkins, Tunnel Establishment Protocol", draft-ietf-mobileipcalhoun-tep-01.txt, Mar. 1998. P. R. Calhoun, and C. E. Perkins, Mobile IP Dynamic Home Address Allocation Extensions", draft-ietfmobileip-home-addr-alloc-00.txt, Nov. 1998. P. R. Calhoun, G. Montenegro, and C. E. Perkins, Mobile IP Regionalized Tunnel Management", draftietf-mobileip-reg-tunnel-00.txt, Nov. 1998. P. R. Calhoun, and C. E. Perkins, Mobile IP Challenge Response Extensions", draft-ietf-mobileip-challenge01.txt, Feb. 1999. P. R. Calhoun, and C. E. Perkins, Mobile IP Network Address Identi er Extension", draft-ietf-mobileipmn-nai-00.txt, Feb. 1999. D. Cong , M. Hamlen, and C. E. Perkins, The De nitions of Managed Objects for IP Mobility Support", RFC 2006, Oct. 1996. A. Conta, and S. Deering, Internet Control Message Protocol ICMPv6 for the Internet Protocol Version 6 IPv6 Speci cation", RFC 2463, Dec. 1998. B. P. Crow, I. Widjaja, J. G. Kim, and P. T. Sakai, IEEE 802.11 Wireless Local Area Networks", IEEE Communications Magazine, Sept. 1997. M. Daniele, IP Version 6 Management Information Base for the Transmission Control Protocol", RFC 2452, Dec. 1998. M. Daniele, IP Version 6 Management Information Base for the User Datagram Protocol", RFC 2454, Dec. 1998. S. Deering, and R. Hinden, Internet Protocol, Version 6 IPv6 Speci cation", RFC 2460, Dec. 1998. M. Dell'Abate, M.D. Marco and V. Trecordi, Performance Evaluation of Mobile IP Protocols in a Wireless Environment", Proc. IEEE ICC, Atlanta, USA, June 1998. 29

Z. J. Haas and P. Agrawal, Mobile-TCP: An Asymmetric Transport Protocol Design for Mobile Systems", Proc. IEEE ICC, Montreal, Canada, pp. 1054-1058, June 1997. S. Hanks, D. Farinacci, and P. Traina, Generic Routing Encapsulation", RFC 1701, Oct. 1994. J. S. Hansen, T. Reich, B. Andersen, and E. Jul, Dynamic Adaption of Network Connections in Mobile Environments", IEEE Internet Computing, Jan. 1998. D. Haskin, and S. Onishi, Management Information Base for IP Version 6", RFC 2466, Dec. 1998. A. Hills and D. B. Johnson, A Wireless Data Network Infrastructure at Carnegie Mellon University", IEEE Personal Commu., Feb. 1996. R. Hinden, and S. Deering, IP Version 6 Addressing Architecture", RFC 2373, Jul. 1998. R. Hinden, M. O'Dell, and S. Deering, An IPv6 Aggregatable Global Unicast Address Format", RFC 2374, Jul. 1998. R. Hinden, and S. Deering, IPv6 Multicast Address Assignments", RFC 2375, Jul. 1998. C. Huitema, IPv6: The New Internet Protocol, 2d ed., Prentice Hall, Upper Saddle River, NJ, 1997. R. Jain, T. Raleigh, C. Gra and M. Bereschinsky, Mobile Internet Access and QoS Guarantees Using Mobile IP and RSVP with Location Registers", IEEE ICC, Atlanta, USA, June 1998. W. Jake, Microwave Mobile Communications, IEEE Press, 1994. D. B. Johnson Route Optimization in Mobile IP", draft-ietf-mobileip-optim-07.txt, Nov, 1997. D. B. Johnson and D. A. Maltz, Protocols for Adaptive Wireless and Mobile Networking", IEEE Personal Commu., Feb. 1996. P. Karn, P. Metzger, and W. Simpson, The ESP DES-CBC Transform", RFC 1829, Aug. 1995. D. C. Lee, D. L. Lough, S. F. Midki , N. J. Davis, and P. E. Bencho , The Next Generation of the Internet: Aspects of the Internet Protocol Version 6", IEEE Network, Jan. Feb. 1998.

30

P. Loshin, TCP IP for Everyone, AP Professional Chestnut Hill, MA, 1995. P. Manzoni and G. Serazzi, Impact of Mobility on TCP IP: An Integrated Performance Study", IEEE JSAC, June 1995. P. Metzger, and W. Simpson, IP Authentication using Keyed MD5", RFC 1828, Aug. 1995. Mobile IP Resources", http: www neda.com mobileIpSurvey html mobileIP.html, Feb. 1998. G. Montenegro and V. Gupta, Firewall Support for Mobile IP", draft-montenegro- rewall-sup-03.txt, Jan, 1998. G. Montenegro, Reverse Tunneling for Mobile IP", RFC 2344, May 1998. G. Montenegro, and V. Gupta, Sun's SKIP Firewall Traversal for Mobile IP", RFC 2356, June 1998. M. Naghshineh, M. Schwartz, and A. S. Acampora, Issues in Wireless Access Broadband networks", in Wireless Information Networks: Architecture, Resource Management, and Mobile Data, edited by J. M. Holtzman, Kluwer Academic Publishers, 1996. M. Naghshineh and M. Willebeek-LeMair, End-to-End QoS provisioning in Multimedia Wireless Mobile Networks Using an Adaptive Framework", IEEE Commu. Magazine, Nov. 1997. T. Narten, E. Nordmark, and W. Simpson, Neighbor Discovery for IP Version 6", RFC 2461, Dec. 1998. C. E. Perkins, and P. Bhagwat, A Mobile Networking System based on Internet Protocol", IEEE Personal Communications First Quarter 1994. C. E. Perkins, IP Mobility Support", RFC 2002, Oct. 1996. C. E. Perkins, IP Encapsulation within IP", RFC 2003, Oct. 1996. C. E. Perkins, Minimal Encapsulation within IP", RFC 2004, Oct. 1996. C. E. Perkins, IP Mobility Support version 2", draft-ietf-mobileip-v2-00.txt, Nov. 1997. C. E. Perkins, Mobile IP", IEEE Communication Maganize, May 1997. 31

C. E. Perkins, Registration Keys for Route Optimization", draft-ietf-mobileip-regkey-00.txt, Nov. 1997. C. E. Perkins, Special Tunnels for Mobile IP", draft-ietf-mobileip-spectun-00.txt, Nov, 1997. C. E. Perkins, Mobile Networking Through Mobile IP", IEEE Internet Computing, Jan. Feb. 1998. C. E. Perkins, Mobile Support in IPv6", draft-ietf-mobileip-ipv6-07.txt, Nov. 1998. Y. Rekhter, and T. Li, An Architecture for IPv6 Unicast Address Allocation", RFC 1887, Dec. 1995. L. A. Sanchez, G. D. Troxel, Rapid Authentication for Mobile IP", draft-ietf-mobileip-ra-00.txt, Nov. 1997. U. Schoen, J. Hamann, A. Jugel, H. Kurzawa, and C. Schmidt, Convergence Between Public Switching and the Internet", IEEE Communications Magazine, Jan. 1998. W. Simpson, IP in IP Tunneling", RFC 1853, Oct. 1995. J. E. Smith and F. W. Weingarten, Research Challenges for the Next Generation Internet", Computing Research Association, May 1997. J. D. Solomon, Applicability Statement for IP Mobility Support", RFC 2005, Oct. 1996. J. D. Solomom, Mobile IP - the Internet Unplugged, Prentice Hall, Upper Saddle River, NJ, 1998. W. Stallings, IPv6: The New Internet Protocol", IEEE Communication Magazine, July 1996. W. Stevens, TCP IP Illustrated, Addison Wesley, Reading, MA, 1994. W. Stevens, TCP Slow Start, Congestion Avoidance, Fast Retransmit, and Fast Recovery Algorithms", RFC 2001, Jan. 1997. M. S. Taylar, W. Wung and M. Banan, Internetwork Mobility - The CDPD approach, Prentice Hall, Upper Saddle River, NJ, 1996. S. Thomson, and C. Huitema, DNS Extensions to support IP version 6", RFC 1886, Dec. 1995. S. Thomson, and T. Narten, IPv6 Stateless Address Autocon guration", RFC 2462, Dec. 1998. 32

P.P. White, RSVP and Integrated Services in the Internet: A Tutorial", IEEE Communications Magazine, May 1997. R. Yavatkar and N. Bhagawat, Improving End-to-End Performance of TCP over Mobile Internetworks", Proc. IEEE'94 Workshop on Mobile Computing Systems and Applications, Santa Cruz, CA, 1994. J. K. Zao, and M. Condell, Use of IPSec in Mobile IP", draft-ietf-mobileip-ipsec-use-00.txt, Nov. 1997.

33