IRED: An Improved Active Queue Management Algorithm

Quality of Service (QoS) of Internet routers is still challenged with the issue of congestion. Active Queue Management (AQM) algorithms aimed at improving end-to-end delay of packets by keeping the average queue size small. This objective is yet to be fully accomplished, especially for interactive services. In this study, an AQM algorithm named Improved Random Early Detection (IRED) algorithm based on the popular Random Early Detection (RED) is presented. IRED deploys a combination of a nonlinear and a linear packet dropping functions. Ns-3 simulation performance evaluations proved that IRED effectively controls the average queue size and delay under light and heavy network traffic conditions. Replacing/upgrading the RED algorithm implementation in Internet routers (either software or hardware) requires minimal effort since only the packet dropping probability profile needs to be adjusted.