Statistical Methods to Estimate the Intrinsic Hurst Parameter From Self-similar Aggregated Data Traffic Of Current High Speed Networks

摘要

The self-similar (fractal) nature of current data networks traffic has been and stills very convincing in the recent past years. High quality analysis of huge packet data time series from LAN and WAN show in a categorical way that the statistical behavior is far away to being a Markovian-type traffic. This paper shows many statistical methods to estimate the Hurst parameter intrinsic from real Internet data traffic (LAN/WAN) gathered from our campus network gateway.