Fitting Algorithms for MMPP ATM Traffic Models

In this paper, we propose and study fitting algorithms for MMPP(2) and CMPP ATM traffic models, which are special cases of Markov Modulated Poisson Processes (MMPP). Two fitting algorithms, both based on the cell interarrival times, are considered for the MMPP(2) model: one fits the cumulative distribution and auto-covariance functions and other fits the first three moments and auto-covariance function. The fitting algorithm for the CMPP model is based on the cumulative distribution and autocovariance functions of the arrival rate. The MMPP(2) is evaluated as a model for the superposition of IPP sources; the CMPP is evaluated as a model for MMPP(2), MMPP(3), MMPP(5), IPP, IDP, Pareto and Self-Similar traffic. The proposed algorithms can be used in the characterisation of ATM traffic streams and in connection admission control procedures.