VBR traffic prediction in ATM system

This paper presents new ATM (asynchronous transfer mode) VBR (variable bit rate) traffic prediction using an artificial neural network (ANN). The paper describes a model of the neural network and its training algorithm. It discusses a new method of designing the neural network using genetic algorithms and it explains the mechanism of three main genetic operations: selection, crossover and mutation. Computer simulation results are presented. It is shown that the neural network can predict the VBR traffic by learning the relationship between ATM traffic variations.