MOBILE IP AND RSVP

Abstract

This thesis looks at how the Mobile IP and RSVP protocols can be used to provide resource reservations for mobile hosts in the Internet. There are a few problems that need to be resolved before Mobile IP and RSVP could be used for this purpose. Firstly, Mobile IP and RSVP do not inter-operate properly with each other. Mobile IP employs IP-in-IP tunnels for data forwarding. However, the basic RSVP protocol is not able to support resource reservation signalling inside tunnels. Secondly, the Mobile IP protocol has not been designed to work well in an environment with frequent handoffs and has poor handoff latency performance. We propose a scheme to provide resource reservations for mobile hosts in the Internet by using enhanced versions of Mobile IP and RSVP, The scheme is simple, modular and flexible by design and is able to support both the portable and wireless picocell environments. We solve the first problem by using the Tunnel Support for RSVP extension, which enables resource reservations to be made inside tunnels. In addition, signalling between the Mobile IP and RSVP processes is also provided by a Mobile Middleware layer to enable RSVP to respond properly to mobility changes. These changes are sufficient to provide resource reservations for mobile hosts in a portable environment. We improve the handoff performance of the Mobile IP protocol by introducing two extensions. The Fast Handoff algorithm and the Regional Aware Foreign Agent extension are used to improve the handoff performance of Mobile IP. These changes improve the performance of our scheme in a wireless picocell environment. We have also investigated some security and scalability issues and solutions with regards to our scheme. In addition, a framework that allows the call dropping probability of a flow in the Internet to be improved has been devised. This works by limiting the number of reserved flows allowed within the administrative domain of a wireless cellular network. We have implemented a subset of our scheme on the Linux operating system as a proof-of-concept. Experimental results based on our implementation have brought up a few issues. We have observed the need to give priority to control packets for these protocols to work properly in a QoS enabled Internet. We have also looked at how to resolve the issue of packet fragmentation, which affects the proper classification of flows.