Performance analysis of delay-loss priority mechanism using Markov modulated arrival stream

This paper proposes a new mechanism to handle the delay and loss requirements of ATM traffic. It is designed to satisfy the different levels of cell loss performance for the two classes of heterogeneous nonreal-time ATM traffic as well as the delay and loss requirements of real-time traffic. The cell arrivals of input streams are modeled by Markov modulated Poisson processes. Both a real-time buffer and a nonreal-time buffer are served by a common server. The buffer occupancies of each traffic are controlled by the partial buffer sharing scheme. The steady state probabilities of the queue length distributions are obtained using the stochastic integral approach. Numerical examples are shown to confirm the effectiveness of the proposed delay-loss priority mechanism.