Measurement-based effective bandwidth estimation for long range dependent traffic

Abstract

Long-range dependent (LRD) or self-similar traffic is becoming more and more popular in today's networks, for example, web traffic, compressed video traffic are all LRD traffic. To provide a satisfactory quality of service (QoS) in a network, the LRD property of sources needs to be considered. However, there are relatively few research publications in regards to the estimation of effective bandwidth and the implementation of a call admission control (CAC) procedure for LRD traffic. This paper approaches the effective bandwidth question from the LRD property of input traffic, and offers a new method for estimating the effective bandwidth for LRD traffic. Our simulation results show that our effective bandwidth estimator is able to effectively bound the overflow probability (or cell loss ratio) of LRD sources and provide the data necessary for a CAC algorithm to effectively manage the call admission control of LRD sources.