A Pareto-modulated Poisson process (PMPP) model for long-range dependent traffic

A new model for long-range dependent traffic is presented. This model is ba sed on a switched Poisson process with two states and the sojourn time in e ach state follows an independent and identical Pareto distribution. It is v ersatile in capturing the self-similar characteristics of traffic found in the recent measurements. Its Index of Dispersion for Counts (LDC) is derive d. It is shown that the LDC increases monotonically and the "log[IDC(t)] ve rsus log[t]" curve has a slope of (2H - 1), where H is the Hurst parameter. Example of the representation of measured data traffic using the introduce d model is given. Performance of a Pareto-modulated Poisson process PMPP/D/ 1 queue is discussed. (C) 2000 Elsevier Science B.V. All rights reserved.