RED-based admission control algorithm for flow-aware networks

Abstract

Measurement-based admission control mechanism in Flow-Aware Networks (FAN) may lead to performance issues related to over-admitting, especially when traffic sources generate large number of flows. When too many new transmissions arrive close in time to each other, all of them may be accepted even though there is no room in the outgoing link. The source lies, unfortunately, in the very basic principles of the mechanism. In this paper, we show that there is a simple, yet very efficient remedy to the problem. The proposed solution utilizes the Random Early Detection algorithm. As a result we can steer the number of accepted flows taking into consideration current load in a link. The proposed solution has significant advantages over currently available algorithms for admission control in FAN. Moreover it provides better performance which is confirmed by the simulation results.