Asymptotic Delay Distribution and Burst Size Impact on a Network Node Driven by Self-similar Traffic

It was shown recently that under self-similar traffic the delay distribution function can decrease very slowly, so in order to guaranty the Quality of Service (QoS) in communication networks, burst size is usually bounded by some value using, for example, leaky-bucket mechanism. In this paper we consider a discrete-time queue with M types of independent input processes. Each input process is the aggregation of sessions (bursts) arrived by a Poisson process. Asymptotic delay distribution at network node driven by self-similar traffic and its effects on burst size bound have been analysed. It is also found the critical value of the burst size at which delays start to increase considerably.