Estimate the loss and tail probabilities for multimedia communication systems

In this paper, we study the performance of integrated services ATM multiplexers in terms of their loss performance. We model voice sources as ON-OFF processes, video sources with uniform activity level as a birth-death process and the arrival process from the aggregation of data sources as a Poisson process with a general bulk size distribution. We show that the integration of these processes belongs to the family of batch Markovian arrival processes (BMAP). We investigate an approximation which matches the original model to a two-state batch arrival Markov modulated Poisson process (BMMPP). We investigate the queue length distribution with particular emphasis on tail probabilities. We develop a fast solution for the asymptotic tail based on the Z-transform which is a very good approximation. The work also provides a simple way to calculate the loss probability for correlated arrivals. The results show that the proposed model has greatly reduced complexity while retaining reasonably accurate performance prediction.