A high performance fuzzy traffic controller for ATM networks

Abstract

In asynchronous transfer mode (ATM) networks, to avoid congestion situations which decrease the quality of service (QoS) of already established connections, usage parameter control (UPC) is necessary. An ideal UPC mechanism must be able to detect any possible violation from the traffic contract. In this paper we use the benefits of fuzzy logic to design a high performance fuzzy traffic controller. In our proposed fuzzy traffic controller, the actual mean cell rate of the traffic source is estimated and the traffic controller is adjusted so its loss load is equal to the generated excessive load. To improve the channel utilization, the proposed fuzzy traffic controller uses feedback from the network to decide whether to pass, mark or discard the input cells. Simulation results show that the proposed fuzzy traffic controller can outperform the traditional UPC mechanisms. It has been also observed that our proposed fuzzy traffic controller improves the channel utilization and demonstrates a low cell loss probability.