IMPROVED EXPLICIT FORWARD CONGESTION INDICATION SCHEME IN ATM NETWORKS USING SIMULATION PROCEDURE

Abstract

Congestion occurs during aggregate resource demand, usually bound by bandwidth, that exceeds usable space. Congestion is especially undesirable, since it results in a longer queue in a shorter period, cell depletion and minimized throughput. Congestion management systems are required to control the flow of traffic in the network to prevent loss of performance in the network. Congestion Management (CM) mechanisms in ATM networks have mainly been categorized as credit-based and rate-based mechanisms. There are several reasons for the inability to incorporate credit-based systems, as it is constrained to reserve capital in advance for any virtual connection and involves a high transmission delay. On the other hand, rate-based procedures are typical because they have an end-to-end feedback system capable of dynamically regulating the transmission rate of each end device source. In this analysis, we considered some common rate-based congestion management mechanisms for ABR traffic, usually implemented in ATM networks, which reveal the deficiencies of each of them. We simulated a simple EFCI system using single bit input from ATM networks. The performance evaluation of the EFCI scheme is focused on the consideration of two parameters, such as the percentage of cell drop at the intermediate switch and the allowable cell rate of the source end device. The simulation results showed that the EFCI system had a higher level of cell drop and enabled the source end system cell rate to oscillate between the peak rate and the minimum rate. We have made an improvement to the basic EFCI or Enhanced EFCI (EEFCI) to solve the problems of oscillation in the approved cell rate and high cell decrease.