Loss probability approximation of a statistical multiplexer and its application to call admission control in high-speed networks

This paper studies the cell loss probability approximation of a statistical multiplexer in high speed networks. For this purpose, we consider discrete-time finite-buffer queueing models. We first propose a simple approximate formula for the cell loss probability in terms of the tail distribution of the queue length in the corresponding infinite-buffer queue. Since the tail distribution has a simple asymptotic expression in many situations, we revisit the asymptotic analysis of the tail distribution. Furthermore we consider specific source traffic models which are particularly important in practice. The formula allows heterogeneous sources and each source is characterized only by a small number of parameters so that they meet engineering requirements. Combining those results, we propose a new call admission control (CAC) scheme which is based on the the cell loss probability as a measure of quality of services.