10.3 Mobility and the Internet Protocols

10.3 Mobility and the Internet Protocols

The Internet was born in an era when no mobile networking equipment was available. Therefore, all the basic protocols were designed under the tacit assumption that the end-points of a communication would stay fixed all along a session. With the arrival of modern communications equipment that allows these end-points to change their position, new protocols for handling mobility have been proposed.

Mobile IP allows mobility of devices, potentially around the world; this is why the type of mobility support it provides is sometimes referred to as global mobility. However, as we will see, mobility within a limited geographical area (called micromobility), has different characteristics and requirements that pose the need for specialized support.

10.3.1 The Problem of IP-Based Mobility

Although networking-enabled mobile devices are becoming more common everyday, most networking protocols — including the TCP/IP protocol suite — have been designed assuming that hosts are always attached to the network at a single physical location. Therefore, host mobility is seen as a rarely occurring fact that can be handled manually. Consider for instance the following scenario: a business executive is usually connected to the network in the office, but occasionally needs to use a laptop computer for meetings; the meeting facilities may be elsewhere in the building or perhaps in a different building or city. If the executive's desk and the meeting room have direct access to the same IP subnet, then the mobility process is trivial. In situations where this is not the case, the only solution is for the user to acquire a new IP address from the appropriate local authority. Then, several configuration files on the moving machine, on various name servers, and on other machines that use the original IP address to identify the moving machine need to be modified. Thus, moving the computer from one place to another involves a slow, error-prone, manual procedure that a typical user does not have the skills or the inclination to deal with. Moreover, even if the process is successfully performed, the mobile host will lose its former identity and will usually need rebooting.

The situation is that, given TCP/IP's early design assumptions that end systems are stationary, if during an active connection one end system moves, then the whole connection breaks, obviously disrupting all networking services layered on top of TCP/IP. Evidence has been given ^[8] that in order to retain transport layer connections, a mobile host's address must be preserved regardless of its point of attachment to the network. The problem with a transport layer protocol such as TCP is that a TCP connection is identified by a 4-tuple:

    <src IP address, src TCP port, dest IP address, dest      TCP port> 

So, if neither host moves, all elements of the tuple remain fixed and the TCP connection can be preserved. However, if either end of the connection moves, the following problem will take place:

• If the mobile host acquires a new IP address, then its associated TCP connection identifier also changes. This causes all TCP connections involving the mobile host to break.

• If the mobile retains its address, then the routing system cannot forward packets to its new location.

These problems come from the very design of IP which, in addition to fragmentation and reassembly, is responsible for "providing the functions necessary to deliver a package of bits (an Internet datagram) from a source to a destination over an interconnected system of networks." ^[9] So, this definition designates responsibility to IP for routing datagrams to and from mobile hosts transparently to higher layers. The problem is that IP addresses serve a dual purpose, as they are not only used by higher layers to identify source and destination hosts, but also by their division into network and host parts which contain location information. Therefore, in its role as an identifier, an IP address must be constant during mobility to avoid affecting higher layers.

Research studies on IP mobility have suggested that mobility is essentially an address translation problem and is best resolved at the network layer. ^[10] As Figure 10.8 shows, a mobile host MH can move away from its home network and attach to the Internet through a foreign network. While away, MH obtains a forwarding address derived from the address space of the foreign network. However, if another host S tries to send packets to MH, it will do so using MH's home address. The problem is resolved by the use of an address translation agent (ATA) at the home network, and a forwarding agent (FA) at the foreign network. These agents perform functions f and g, respectively, which are defined as follows:

• f: home address → forwarding address

• g: forwarding address → home address

Figure 10.8: Mobility as an address translation problem.

This way, when S sends packets to MH, they first pass through ATA. This agent performs mapping f to send the packets to the address that MH acquired in the foreign network. At the foreign network, FA intercepts all packets containing MH's forwarding address. It then proceeds to apply the function g to map from this forwarding address to MH's original home address and effectively forward the packets.

10.3.2 Mobile IP

In order to react to the new challenges posed to the Internet architecture by the arrival of mobile networking devices, the IETF created the Mobile IP Working Group. The basic Mobile IP standard ^[11] specifies a mobility management architecture for the Internet. In principle, both local-area and wide-area mobility across wired and wireless networks can be handled, although certain inefficiencies have been detected. Later, we will see extensions to Mobile IP proposed to overcome such inefficiencies.

Figure 10.9 shows the basic operation of Mobile IP. A mobile node is normally attached to its home network using a static home address. When the mobile node moves to a foreign network, it makes its presence known by registering with a foreign agent (FA). The mobile node then communicates with a home agent (HA) in its home network, giving it the care-of address (COA), which identifies the foreign agent's location. Typically, routers in a network will implement the roles of home and foreign agents. When IP datagrams are exchanged over a connection between the mobile node A and a correspondent host B, the following operations occur: ^[12]

1. Host B transmits an IP datagram destined for mobile node A, with A's home address in the IP header. The IP datagram is routed to A's home network.

2. At the home network, the incoming IP datagram is intercepted by the home agent. The home agent encapsulates the entire datagram inside a new IP datagram, which has A's care-of address in the header, and retransmits the datagram. The use of an outer IP datagram with a different destination IP address is known as tunneling.

3. The foreign agent strips off the outer IP header, encapsulates the original IP datagram in a MAC-level PDU (for example, an Ethernet frame), and delivers the original datagram to A across the foreign network.

4. When A sends IP traffic to B, it uses B's IP address. In our example, this is a fixed address; i.e., B is not a mobile node. Each IP datagram is sent by A to a router on the foreign network for routing to B.

5. The IP datagram from A to B travels directly across the Internet to B, using B's IP address.

Figure 10.9: Basic mobile IP scenario.

10.3.4 Mobile IP Problems

There are currently several outstanding problems facing Mobile IP, posing technical as well as practical obstacles for its deployment. ^[13] One of the most notable problems is due to routing inefficiencies. In the basic Mobile IP protocol, IP packets destined to a mobile node (MN) that is outside its home network are routed through the home agent. However, packets from the MN to the corresponding nodes are routed directly. This is known as triangle routing (see Figure 10.10).

Figure 10.10: Triangular routing.

This method may be inefficient when the correspondent host and the MN are in the same network, but not in the same home network of the MN. In such a case, the messages will experience unnecessary delay because they have to be routed first to the HA that resides in the home network. In order to alleviate this, a technique known as route optimization has been proposed. ^[14] However, implementing it requires changes in the correspondent nodes that will take a long time to deploy in IPv4.

Some other problems are related to performance and scaling issues. Studies have shown that Mobile IP can suffer from unacceptably long handoff latencies when the mobile host is far from its home network. ^[15] Scalability can be a problem as the number of mobile hosts grow, but in this case the network is the bottleneck, as mobility agents (i.e., HAs, FAs) can easily service at least a few hundred hosts. Suggestions have been made that using a hierarchical model to manage mobility could reduce or eliminate these performance and scaling problems. ^[16], ^[17] Security is also a particular area of attention in Mobile IP.

A lot of the problems of Mobile IP are related to the lack of features for streamlining mobility support in IPv4. ^[18] Some of these problems may be solved by IPv6. While Mobile IP was originally designed for IPv4, IPv6 ^[19] incorporates features that support mobility much more easily; several mechanisms that had to be specified separately now come integrated with IPv6. Some of these IPv6 features include stateless address autoconfiguration ^[20] and neighbor discovery. ^[21] IPv6 also attempts to drastically simplify the process of renumbering, which may be critical to the future of routability of the Internet. ^[22] Security is a required feature for all IPv6 nodes.

10.3.5 Micro-Mobility

As several studies indicate, ^[23], ^[24] users' mobility patterns are highly localized. For instance, business professionals may spend a considerable amount of time away from their desks, but once away, most of their mobility will take place within the same building. While the mobile user is at the foreign administrative domain, there is no need to expose motion within that domain to the home agent or to correspondent hosts in other domains. Therefore, mobility management within an administrative domain should be separate from global mobility management.

In principle, Mobile IP can handle both global and local mobility. However, it requires that the mobile's home network is notified of every change in location. Moreover, route optimization extensions ^[25] further require that every new location is registered with hosts that are actively communicating with the mobile node. All these location updates incur communications latency and also add traffic to the wide-area portion of the internetwork. Therefore, Mobile IP does not extend well to large numbers of portable devices moving frequently between small cells. It also has been demonstrated that, when used for micro-mobility support, Mobile IP incurs disruption to user traffic during handoff, and high control overhead due to frequent notifications to the home agent. ^[26] Another type of protocol, a micro-mobility protocol, ^[27] is then needed for local environments where mobile hosts change their point of attachment to the network so frequently that the basic Mobile IP tunneling mechanism introduces network overhead in terms of increased delay, packet loss, and signaling.

Acknowledging the fact that Mobile IP may not be the universal end-all solution for mobility on the Internet, its performance and scalability challenges have been under discussion. Within this context, the Mobile IP working group has recently started discussing the subject of micro-mobility protocols. There are several attributes that micro-mobility protocols aim toward:

• Minimum (or zero) packet loss: Fast handoff techniques have been developed to achieve this, and they may reduce latency or delay.

• Reduced signaling: Techniques for locating mobile hosts, known as paging, have been proposed in order to reduce signaling. Reduced registration is also an outcome of these techniques.

HAWAII and Cellular IP are two prominent proposals for micro-mobility management and we give a brief presentation of both:

• HAWAII ^[28], ^[29] (Handoff-Aware Wireless Access Internet Infrastructure) is an alternative for providing domain-based mobility (i.e., micro-mobility). Under this approach, Mobile IP is used as the basis for mobility management in wide-area wireless networks, but new methods for managing mobility within an administrative domain are developed. One point worth highlighting is that mobile hosts retain their network address while moving within a domain; this way, the Home Agent (HA) — if using Mobile IP — and any corresponding hosts are not aware that the host has performed intradomain mobility. Dividing the network into hierarchies, loosely modeling the autonomous system hierarchy used in the Internet, is part of the HAWAII approach. Indeed, the gateway into each domain is called the domain root router, and each host is assumed to have an IP address and a home domain. As already stated, hosts retain their address while moving within a domain, so when packets destined to a mobile host arrive at the domain root router, they are forwarded over specially established paths to reach the mobile host. However, if the mobile host moves to a foreign domain, traditional Mobile IP mechanisms are used.

• Cellular IP ^[30], ^[31] aims to integrate cellular technology principles with the IP networking paradigm; this poses difficult challenges, as there are fundamental architectural differences between cellular and IP networks. A Cellular IP node constitutes the universal component of a Cellular IP network, because it serves as a wireless access point but at the same time routes IP packets and integrates cellular control functionality traditionally found in mobile switching centers (MSC) and base station controllers (BSC). Cellular IP nodes are modified IP nodes where standard routing is replaced by Cellular IP's own routing and location-management functions. A Cellular IP network is connected to the Internet via a gateway router. Mobility between gateways (i.e., Cellular IP access networks) is managed by Mobile IP, while mobility within access networks is handled by Cellular IP. Mobile hosts attached to the network use the IP address of the gateway as their Mobile IP care-of address.

Another important aspect that has received little attention in the design of micro-mobility protocols is that of quality of service (QoS). Triangular routing, address translation, and complex interaction between agents make Mobile IP unsuitable for QoS support in local environments. ^[32], ^[33], ^[34]

^[8]Bhagwat, P., Perkins, C., and Tripathi, S., Network layer mobility: an architecture and survey, IEEE Personal Communications Magazine, 3, 54, 1996.

^[9]DARPA, DARPA Internet Program Protocol Specification, Internet RFC 791, 1981.

^[10]Bhagwat, P., Perkins, C., and Tripathi, S., Network layer mobility: an architecture and survey, IEEE Personal Communications Magazine, 3, 54, 1996.

^[11]Perkins, C., Mobile IP specification, Internet RFC 2002, 1996.

^[12]Stallings, W., Mobile IP, The Internet Protocol Journal, 4, 2, 2001.

^[13]Chesire, S. and Baker, M., Internet mobility 4x4, in ACM SIGCOMM Computer Communications Review, 318, 1994.

^[14]Johnson, D. and Perkins, C., Route optimization in mobile IP, IETF Mobile-IP draft, July 1995.

^[15]Mukkamalla S. and Raman, B., Latency and scaling issues in mobile IP, Iceberg Project technical report, University of California, Berkeley, 2001.

^[16]Caceres, R. and Padmanabhan, V., Fast and scalable handoffs for wireless internetworks, in ACM Mobicom 96, 1996.

^[17]Soliman, H. et al., Hierarchical MIPv6 mobility management (HMIPv6), Internet draft draft-ietf-mobileip-hmipv6-04.txt, work in progress, July 2001.

^[18]Perkins, C., Mobile networking through mobile IP, IEEE Internet Computing, 2 (1), 1998.

^[19]Deering, S. and Hinden, R., Internet Protocol version 6 (IPv6), Internet RFC 1883, 1995.

^[20]Thomson, S. and Narten, T., IPv6 stateless address autoconfiguration, Internet RFC 1971, 1996.

^[21]Narten, T., Nordmark, E., and Simpson, W., Neighbor discovery for IP version 6 (IPv6), Internet RFC 1970, 1996.

^[22]Castineyra, I., Chiappa, J., and Steenstrup, M., The Nimrod routing architecture, Internet RFC 1992, 1996.

^[23]Kirby, G., Locating the user, Communications International, 1995.

^[24]Toh, C., The design and implementation of a hybrid handover protocol for multimedia wireless LANs, in Proc. 1st International Conference on Mobile Computing and Networking, 1995.

^[25]Johnson, D. and Perkins, C., Route optimization in mobile IP, IETF Mobile-IP draft, July 1995.

^[26]Caceres, R. and Padmanabhan, V., Fast and scalable handoffs for wireless internetworks, in ACM Mobicom 96, 1996.

^[27]Campbell, A. and Gomez-Castellanos, J., IP micro-mobility protocols, ACM Sigmobile Mobile Computer and Communications Review, 2001.

^[28]Ramjee, R. et al., IP micro-mobility support using HAWAII, Internet draft draft-ietf-mobileip-hawaii-01.txt, work in progress, July 1999.

^[29]Ramjee, R. et al., HAWAII: a domain-based approach for supporting mobility in wide-area wireless networks, in IEEE International Conference on Network Protocols, 1999.

^[30]Valko, A., Cellular IP — a new approach to Internet host mobility, ACM Computer Communication Review, 1999.

^[31]Campbell, A. et al., An overview of cellular IP, in IEEE Wireless Communications and Networks Conference, WCNC, 1999.

^[32]Chan, J. et al., The challenges of provisioning real-time services in wireless Internet, Telecommunications Journal of Australia, 2000.

^[33]Helal, A. et al., Towards integrating wireless LANs with wireless WANs using mobile IP, in IEEE Wireless Communications and Networks Conference, WCNC, 2000.

^[34]Mukkamalla, S. and Raman, B., Latency and scaling issues in mobile IP, ICEBERG Project technical report, University of California, Berkeley, 2001.