Evaluation of ABR congestion control protocols for ATM LAN and WAN

Abstract

This paper evaluates and compares six rate-based congestion control protocols for the ABR (Available Bit h:ate) trafic over the ATM (Asynchronous Transfer Mode) networks. They include: the EFCI-bit setting, the EFCI-bit setting with separate RM queues, the CI-bit setting in the backward direction, the CIbit setting in the backward direction with separate RM queues, the CAPC2 ER (Explicit Rate) scheme, and the EFCI-bit setting with utilization-based congestion indication. Each scheme is simulated and compared in the LAN, WAN, and GFC (General Fairness Configuration) environments specified b y the ATM Forum. Effects of varying VC (Virtual Circuits) number and changing endsystem-switch distance has been investigated. Their fairness is also compared using the GFC confi:guration. For each simulation run, we measure average queueing delay, maximum queue length, and network utilization. Traces of ACR (Allowed Cell Rate) and buffer queue length are also examined. We have found that the ER control scheme performs significantly better than the other five binary control schemes by its faster response to congestion, smoother regulation of bit rates, lower queueing delay, shorter buffer queue length, and fairness. Among the other five schemes, the CI-bit setting scheme performs better than the EFCI bit setting scheme. Providing separate RM queues has significmtly improved the EFCI scheme in the WAN environment, but has little effect on the CI scheme. Link utilizlation-based congestion detection has sufFered from either low utilization or an excess cell loss which is unacceptable in most data applications. I11 IV IhtroductionIn providing integrated service for multimedia and heterlogeneous traffic, the ATM Forum traffic management specification has categorized five traffic and service classes [l]: constant bit rate (CBR), real-time variable bit rate (rt-VBR), non-real-time variable bit rate CI-bit setting in the backward direction CI-bit setting with separate RM-cell queues