Static vs. adaptive feedback congestion controller for ATM networks

One of the fundamental challenges facing broadband information transport is the determination of congestion control strategies to support multiple classes of traffic in the asynchronous transfer mode (ATM) based networks. Monitoring the buffer status (queue length) is the most commonly used mechanism to detect congestion in ATM networks. In a static feedback controller approach, feedback signals are generated when the queue length of the buffer exceeds a threshold value. However, defining the threshold of the buffer as a congestion point is not straightforward and the value to which source rates must be regulated is not so clear, either. We propose an explicit congestion notification mechanism for ATM networks using artificial neural networks (ANNs) to estimate the amount by which sources need to reduce their transmission rates. Three models using the ANNs are presented and the obtained results are compared. The results of a simulation study suggest that our mechanisms provide simple and effective traffic management for ATM networks. Cell loss due to congestion shows a 2-10 times improvement compared with the static approach, while transmission delay introduced by our ANN controller is also smaller than in the static approach in most cases.