QoS Based Predictive Admission Control for Multi-Media Traffic

Abstract

In this work, we develop a distributed predictive admission control algorithm which provides appropriate throughput to the system while maintaining a fixed call dropping probability defined by the user. This algorithm is then extended to include multi-media traffic where the QoS criteria are the respective call dropping probabilities of the different traffic classes. The traffic classes are completely defined in terms of their individual call service time, handoff, and bandwidth parameters. This work is based on simple Markov analysis and traffic prediction. Using this algorithm, it is possible to combine any number of traffic classes in a network flexibly while maintaining the unique QoS criterion of each class. For the sake of simplicity, the algorithm is developed for a single dimensional wireless network. However, we show how it may be easily extended to multi-dimensional wireless networks.

This work was supported by ARO under Grant DAAH04-95-1-0188