Design of adaptive real queue control algorithm supporting TCP flows

Hao Wang, Ou Li, Chenda Liao, Zuohua Tian

Abstract: Active queue management (AQM) mechanism plays an important role in network congestion control. In this paper, we propose an effective AQM algorithm, named adaptive real queue control (ARQC). In particular, an explicit congestion indicator, virtual regulating time, is designed to detect network congestion. Then by analyzing the queuing system on routers, we present the guideline to calculate the dropping/marking probability. Also we provide the tuning rules for the control parameters based on the discrete model of TCP dynamics. Simulation results demonstrate that ARQC outperforms the other AQM schemes in terms of stability, responsiveness and robustness in dynamic networks. Moreover, it achieves smaller queue length jitter, less packet loss and better trade-off between delay and link utilization.