Performance enhancement of ERICA+ using fuzzy logic prediction

The long propagation delay and the existence of highly bursty VBR traffic pose a significant challenge for any end-to-end feedback control scheme on an ATM network. A fuzzy logic based scheme aims at improving the QoS of the ABR service under these conditions is proposed. This paper applies the proposed fuzzy logic control technique of Qiu(see Proc. IEEE GLOBE-COM'97, Arizona, USA, p.967-71, 1997 and Proc. IEEE ICC'98, Atlanta, Georgia, USA, vol. 3, p.1769-73, 1998) to the standard proposed switch algorithm-Explicit Rate Indication for Congestion Avoidance (ERICA+) so as to study its performance. A fuzzy logic predictor is used to estimate the ABR output switch queue length one round-trip delay in advance. This information, together with the total queue growth rate and current ABR queue length are provided to a fuzzy inference system which generates a rate factor. This factor is used as an additional component to increase/decrease the explicit rate (ER) value of a backward resource management (BRM) cell. Simulation results show that the fuzzy logic control scheme can out-perform ERICA+ in terms of queue delay, variation and cell loss ratio under different propagation delays with the existence of CBR and VBR traffic.