A New Self-Similar Processes Network Traffic Model——New S4 Model
HU Yu-qin · Microcomputer Information · 2008
There is growing evidence that aggregate network traffic traces are statistically self-similar.As a Gaussian self-similar model,FBM model is used to account for the existence of similar statistics in different time scales,but since it is Gaussian process it cannot model burstiness well.Recent experimental studies have shown that actual aggregate traffic is strongly non-Gaussian in many cases.In the paper,a model based on α-stable self-similar processes,which can capture both long-range dependent and burst(heavy-tail) characteristics of the traffic is studied.Based on the deep research of existent theory,the paper proposes a improved model based on standard skewed linear fractional stable noise.The parameters of the model are defined very clearly and have the merit of consistency.