Studies on Congestion Control Mechanisms in the Internet – AIMD-based Window Flow Control Mechanism and Active Queue Management Mechanism

AQM (Active Queue Management) mechanisms support an end-to-end congestion control mechanism of TCP (Transmission Control Protocol). Researchers have claimed several advantages of AQM mechanisms over a conventional Drop-Tail gateway such as the small average queue length (i.e., the number of packets in the buffer). Several AQM mechanisms have been recently proposed and studied by many researchers. One of promising AQM mechanisms is the RED (Random Early Detection) gateway, which randomly discards arriving packets. Although its steady state performance has been fully investigated, its transient behavior has not been well understood. RED randomly drops the arriving packet with a probability proportional to its average queue length. However, it is unclear whether the packet marking function of RED is optimal or not. In this paper, we first analyze the transient behavior of the RED gateway for various types of TCP connections variations. We use a control theoretic approach by utilizing the transfer function, which describes the relation between the input and the output in frequency domain. By presenting several numerical and simulation results, we discuss how control parameters of the RED gateway affect its transfer behavior. We also investigate what type of packet marking function, which determines the packet dropping probability from the average queue length, is suitable from the viewpoint of steady state performance and transient behavior. By presenting several numerical examples, we investigate advantages and disadvantages of three packet marking functions: linear, concave, and convex. We show through numerical examples that although the average queue 1 length in steady state becomes larger, use of the concave function makes the transient behavior of RED and the robustness against network status changes better.