Explicit Fairness Enhancement Mechanism and Its Classifier Implementation in Congestion Control

Active Queue Management (AQM), which can maintain smaller queuing delay and higher throughput by purposefully dropping packets at intermediate nodes, has been intensively studied. Most of the existing AQM schemes focus on the stability, but neglect the fairness. The unfairness caused by the multiple congested gateways and heterogeneous Round Trip Time (RTT) is common in all AQM schemes. In previous works, some researchers developed a framework for AQM scheme based on a considerably simple Two Category Classifier (TCC) [1] instead of the probability discrimination used by most of AQM schemes. In order to improve the fairness, we propose an Explicit Fairness Enhancement (EFE) mechanism by extending TCP header, and then design the TCC-EFE algorithm through adding one more fairness dimension into the two-category classifier to build a three-dimensional classifier. The simulations show that TCC-EFE enhances the fairness.