The Wavelet Detect Of Self-Similarity In The Network
zhi yong guo,jian ping li,jian ming liao,xiao feng gu,si yu zhan
DOI: https://doi.org/10.1142/9789812772763_0135
2006-01-01
Abstract:We don't consider the burst of the traffic stream in traditional network, it is short-range dependence. The recent research has shown the existence of self-similarity and long-range dependence in real network traffic. First of all, we give several familiar definitions of self-similarity (also called long-range dependence), describing the stochastic process's characteristics in mathematics and physics, and then we will discuss how to identify a sequence is self-similarity (long-dependence) or not and how to identify its magnitude in quantity. In this paper, we produce the traffic source with the Pareto distribution (one of the heavy-trail distribution) by the ON/OFF model. Subsequently, we estimate the H parameter using V-T and wavelet methods. As the V-T method, we emphasize on the exactness and the operation time. As the Wavelet method, we analyze the effect that the parameters of wavelet method put on the result and introduce the optimal scheme for choosing the parameters. Base on the optimal scheme, we gain very good result. The result of simulation shows that the estimated parameters correctly describe the characteristics of self-similar traffic stream. Whether exactness or the operation time, Wavelet method performs better. At the same time, we study the effect that the vanishing moment of wavelet put on the result by simulation. We gain the applicable scope of wavelet method.