Statistical multiplexing of identical bursty sources in an ATM network

The authors study the performance of a statistical multiplexer with a common buffer and bursty sources. A uniform arrival and service model has been used to calculate the loss probability as a function of several parameters. It is shown that the ratio of the number of cells in an average burst to buffer capacity per source is a more important parameter than burstiness. When this ratio is much greater than one, performance will be poor unless the load is very low. If the ratio is of the order of one or less, acceptable performance can be achieved. It is shown that combining small load with with multiplexing of many sources can be a very efficient way to lower cell loss probability