Self-similarity and stationarity of increments in VBR video

Abstract

As self-similarity trail is being detected in many types of traffic, and the Markovian models failing to represent some statistical behaviors, the tools being used for traffic testing are still complex. Our study here is related to VBR video. Its self-similarity and long-range dependence aspects will be tested using a wavelet-based tool. As the test tool requires stationarity of the increments of the traces, a novel testing technique will be suggested for this aim. Then, the degree of self-similarity will be related to both the traces time scale and its statistical measures of spreading.