Congestion control in B-ISDN based on a fluid flow dynamic model

We propose a dynamic feedback flow control method for reactive congestion control in B-ISDN (Broadband Integrated Service Digital Network) based on ATM (Asynchronous Transfer Mode) switching and multiplexing technology. Our approach adopts a fluid flow model to represent the dynamic behaviour of the queue length in a bottle-neck link. Based on this model a forecast of the future state of the buffer is used to predict congestion. A feedback control scheme avoids or reduces cell loss by warning the controllable source to reduce its traffic delivery rate. An intelligent controller proposed for the controllable source can intelligently adjust its delivery rate according to feedback information and local delivery history. Comparison of our feedback control scheme with normal open-loop flow control methods using simulation results show that cell loss is controlled over a wide range of traffic flow, and that the simple scheme is robust.<>