Statistical multiplexing of random processes in packet switching networks

The authors present an analytical model for the closed form evaluation of relevant parameters characterising the process at the output of a statistical multiplexer. Compared with previous approaches, the proposed method is able to additionally describe transient conditions and, through the autocorrelation function and the mean length of adjacent packets sequences, to evaluate the extent of the correlation in the output process as a function of that in the input traffic, also taking into account the effect due to a finite buffer size. The reliability of the model is tested, under different operation conditions, by comparison with simulated results, showing everywhere an excellent agreement