Time Scale Analysis of an ATM Queueing System with Long-Range Dependent Traffic

Several types of network trafic have been shown to exhibit long-range dependence (LRD). In this work, we show that the busy period of an ATM system driven by a long-range dependent process can be very large. We introduce a new trafjic model based on a fractional Brownian motion envelope process. We show that this characterization can be used to predict queueing dynamics. Furthermore, we derive a new framework for computing delay bounds in ATM networks based on this trafic model. We show that it agrees with results given by large deviation theory with less computational complexity.