Optimal resource allocation for multi-service networks

Abstract

The next generation of communication networks will simultaneously offer multiple service classes capable of supporting both real-time and best-effort traffic. At the level of an individual network router or gateway, a critical question that needs to be addressed is the development of a set of policies and metrics to determine how router resources should be shared between different service classes. We describe a solution to this problem that combines pricing policies with admission control for real-time traffic. We present a measurement-based approach for adaptive pricing that results in near-optimal resource allocation policies between real-time and best-effort traffic. We also show, through simulation, an application of our approach that maximizes overall user value.