The Influence of Traffic Self-Similarity on QoS Mechanisms

We investigate the influence of stochastic features (self-similarity, long range autocorrelation) of the input traffic on the performance of some QoS mechanisms. We model the impact of input traffic characteristics on queue distributions and packet loss probability observed at leaky bucket mechanism, at fixed priority scheduling of packets and on the RED queue. The results confirm a strong influence of self-similarity on the behaviour of these mechanisms.