Modelling the output process of an ATM multiplexer with Markov modulated arrivals

In ATM networks supporting B-ISDN, the traffic is highly correlated and neglecting its correlation leads to dramatic underestimation of the end-to-end delay and end-to-end loss probability. Being able to accurately estimate end-to-end performance is of paramount importance for traffic control. The purpose of this paper is to introduce a procedure for modelling the output process of a finite buffer discrete-time queue loaded with a Markovian arrival process (D-BMAP/D/1/K). The procedure matches the output process long-term index of dispersion for counts and correlation of the number of arrivals in consecutive slots with the same statistics of a two-state Markov modulated Bernoulli process. We also introduce a framework for the analysis of queueing networks with Markov modulated type of flow.<>