High-speed self-similar ATM VBR traffic generator

Abstract

Recent analysis of real network traffic has shown that a large number of traffic sources produces a traffic stream that is self-similar over several time scales. Motivated by this observation a high-speed Asynchronous Transfer Mode (ATM) traffic generator based on fractional Brownian motion is presented. The traffic simulator generates call arrivals as well as the desired traffic parameters for each call. It is therefore an ideal tool to analyze a wide spectrum of networking functions such as switch performance, routing, and call admission control algorithms. The presented algorithm can be highly parallelized, thus generating traffic logs in a period several orders of magnitude below the actual simulated time period.