Modelling aggregate heterogeneous ATM sources using neural networks

It is proved that artificial neural networks are adequate tools for obtaining superposition models of multiplexed ATM traffic sources. It is shown how complex mathematical models can be replaced by a modular, adaptive and parallel architecture capable of developing complicated algorithms. In particular. the authors approximate a superposition of individual ATM sources by a two-stale Markov modulated Poisson process (MMPP). This approximation is performed using a neural system, matching four statistics of the aggregate traffic to those of the MMPP. The approach is evaluated using numerical examples, showing that it is adequate for estimating delay attributes and cell-level congestion.