A Classified AIMD Congestion Control Scheme in IP Networks

Abstract

At present, TCP congestion control, a reliable rate transmission framework is widely applied to the Internet, by deploying window-based additive increase multiplicative decrease (AIMD) congestion control. However, the congestion control is not favorable for real-time traffic. In this paper, we propose a classified additive increase and multiplicative decrease scheme named C-AIMD and present a prototype in real IP network. Local stability for the proposed scheme is proved by Lyapunov approach, and evaluates its performance both analytically and experimentally in terms of throughput and loss probability among ABR flows and VBR flows in IP networks. Our obtained results have also demonstrated that the C-AIMD scheme can achieve remarkable performance and have higher resilience in IP networks than that of the current TCP framework used.