An ATM call admission control scheme using neural network

Abstract

In ATM networks, it is important to develop a method for allocating network resources efficiently to individual calls by judging admission of calls using different media which require different transfer qualities. The authors propose a new ATM call admission control scheme based on cell transfer state monitoring which does not require that users specify values of traffic descriptors in detail when they originate calls. The authors also propose neural network application to this proposed control scheme in order to resolve a serious problem of this control scheme. Adaptability to unexpected changes in characteristics of traffic applied to the ATM network can be improved by the application of neural network.