Can multifractal traffic burstiness be approximated by Markov modulated Poisson processes?

Internet traffic displays multifractal scaling in local time. In this paper we investigate the approximating capacity of Markov modulated Poisson processes (MMPPs) for modeling multifractal traffic. The choice of MMPP is motivated by that it can capture both the variability and correlation in moderate time scales while it is analytically tractable. Our methodology of doing this is to examine whether the MMPP can be used to predict the performance of a queue to which MMPP sample paths and measured traffic traces are fed for comparison respectively. We also present a customized moment-based fitting procedure of MMPP to traffic trace. Numerical results and simulations show that the fitted MMPP can approximate multifractal traffic quite well, i.e., predict the queueing performance accurately.