ADAPTIVE RED : AN EFFICIENT ACTIVE QUEUE MANAGEMENT SCHEME

Abstract

Congestion has long been a problem in computer networks. During the past decades, much effort has been devoted to understanding the nature of congestion and developing techniques for its control. Most of the earlier proposed congestion control schemes assume the cooperation of network users, requiring them to implement a particular flow control at the end hosts. Unlike flow control, which is concerned with the source not overrunning the receiver, congestion control is about keeping the source from over loading the network. In this context, most of focus, -has been on mechanisms targetted at the active queue management schemes of today's Internet. An important requirement is to prevent congestion collapse, to keep congestion levels low, to keep low packet-loss rates and to achieve high link utilizations. However, even these enhancements, TCP connections still experience high loss rates, especially during times of congestion. The RED active queue management algorithm allows network operators to achieve high throughput and low average .delay. While RED can potentially improve packet loss rates, its effectiveness is highly dependent on its operating parameters. In fact, in cases where these parameters do not match the requirements of the network load, the performance of the RED gateway can approach that of a traditional drop-tail gateway. In this report, to alleviate this problem, a self-configuring queue management scheme Adaptive RED is discussed which can significantly reduce packet loss rates across congested links while maintaining high link utilizations. I have used network simulator ns-2 for performing the experiments for RED and Adaptive RED.