Self-similar traffic analysis based on the M/G/∞ input processes

Abstract

In recent years many studies have shown that network traffic exhibits burstiness on a wide range of time scales with properties of self-similarity. The measurements of network traffic collectively revealed that self-similar and long-range dependence phenomena widely exist in network traffic, and those phenomena can have significant impact on network performance. In this study, we develop a self-similar traffic generator based on a discrete time version of the M/G/∞ input process. Then some properties of the self-similar model are extracted. Throughout experiment, we verify that our self-similar traffic generator performs well.