Markov-modulated self-similar processes: MPEG coded video traffic modeler and synthesizer

Markov modulated self-similar processes are proposed to model MPEG video sequences that can capture the LRD (long range dependency) characteristics of video ACF (auto-correlation function). An MPEG compressed video sequence is decomposed into three parts according to different motion/change complexity such that each part can individually be described by a self-similar process. Beta distribution is used to characterize the marginal cumulative distribution function (CDF) of each self-similar processes, and Markov chain is used to govern the transition among these three self-similar processes. Network cell loss rate using our proposed synthesized traffic is found to be comparable with that using empirical data as the source traffic