Self-Tuning Price-Based Congestion Control Supporting TCP Networks

Active queue management (AQM) is an effective method acting on routers to improve the performance of end-to-end congestion control. However, most of the current AQM schemes are sensitive to network configurations as well as parameterization problems. In this paper, we propose a self-tuning congestion control scheme, named SPC. An enhanced price with proportional-integral-derivative control property is introduced into SPC to improve its capability of detecting and controlling network congestion. Then, we present a simple and effective discrete model to depict the TCP/SPC behaviors, and design an optimization rule for the key parameter of SPC to minimize the error and jitter of queue size. Simulation results in NS2 demonstrate that SPC is stable, responsive and robust in various network scenarios. It can quickly regulate the queue length to the target with small jitter, achieving satisfactory performance, such as low packet loss ratio and high link utilization.