A MODEL OF PACKET TRAFFIC USING A RANDOM WALL MODEL

Abstract

We consider the problem of modeling self-similar bursty packet traffic using chaotic maps. In particular, we introduce a family of maps to overcome the difficulties of describing the self-similar statistics of the packet traffic as a function of the maps' parameters. The packet traffic that can be modeled with these maps varies from Poisson to bursty traffic with different decay tails in the ON and OFF regions. In contrast with Fractional Brownian Models, these maps have a multifractal spectrum so they can generate packet traffic with the same mean, variance and self-similar parameter but different higher order statistics.