A neural network approach towards adaptive congestion control in broadband ATM networks

Abstract

The authors present an adaptive control scheme based on neural networks to solve a general quality-of-service (QOS) control problem in broadband ATM (asynchronous transfer mode) networks. The control algorithms developed for training neural networks are a direct application of the error backpropagation learning method with those modifications required to pose the problem in a QOS control framework. To illustrate the present scheme's ability to control, examples of dynamic models are studied through simulations.<>