Performance Modelling of Random Early Detection Based Congestion Control for Multi-Class Self-Similar Network Traffic

Active Queue Management (AQM) has been proposed as an efficient policy of congestion control to achieve high system utilization and low packet delay. Many recent high-quality measurement studies have demonstrated that realistic traffic in multimedia communication systems exhibits self-similar nature, which has a significant impact on network performance and user-perceived Quality-of-Service (QoS). Performance modelling of congestion control for self-similar traffic is a challenging and important research issue. In this paper, we present a new analytical model for the AQM congestion control policy based on the well-known Random Early Detection (RED) scheme in the presence of multi-class self-similar traffic. The system buffer is partitioned by a number of thresholds in order to provide differentiated loss priorities to individual traffic classes. We derive the closed-form expressions for calculating the loss probabilities of different traffic classes. By comparing analytical and simulation results, we demonstrate the effectiveness and accuracy of the developed model.