Curvilinear RED: An Improved RED Algorithm for Internet Routers

Samuel Hassan and Ayodeji O. Oluwatope


Active Queue Management, Random Early Detection, Non-linear RED, Congestion Control


Random Early Detection (RED), an active queue management (AQM) scheme has been known to address the problem of network congestion in Internet routers. However, RED suffers from large delay which can be partly traced to the single linear packet drop function it deploys. In this paper, we present a new RED-based AQM scheme called Curvilinear Random Early Detection (CLRED) scheme which modified the single linear dropping function of RED with a two-segment (that is, a quadratic and a linear) dropping functions to address the reported drawback. Simulation carried out in network simulator 3 (NS-3) confirms that the proposed CLRED scheme achieved a significant reduction in the average queue size when compared with the classical RED scheme. Therefore, the existing RED implementations can be upgraded/replaced with the proposed CLRED scheme so as to help RED overcome its large delay drawback.

Important Links:

Go Back