Integrated ATM traffic control by distributed neural networks

Abstract

Adaptive control using neural networks for the integration of service quality control in an asynchronous transfer mode (ATM network is described. The ATM is regarded as a key technology for the broadband integrated services network (B-ISDN). The proposed ATM network controller uses a number of back propagation neural networks in various control functions. It is adoptable to mixtures of multimedia traffic and changing situations, because neural networks learn the exact relationship between the offered traffic and service quality. Also the concurrent training of all neural networks distributed over the ATM network integrates and optimizes the whole network control system. In this paper, the computer simulation results of the adaptive control method applied to a basic call admission control model and an integration of admissian control and link capacity assignment are shown.