Active Queue Management Algorithm Based On Rate And Queue Length

Because of rapid expanse of the internet, the proliferation of users and theapplication of all kinds of multimedia, networks are experiencing an increasing numberof packets losses and other performance degradation. As a result, network congestion hasbecome a bottleneck of the development of the Internet. Therefore, proactive measuresare needed to be taken to prevent congestion from occurring and to satisfy certain QoSrequirements. AQM(Active Queue Management) strategy, configured on theintermediate nodes, is one of the primary means to improve the network performance,and to enhance the quality of services. This strategy has become a hot research area incongestion control in recent years.This paper expounds the development of congestion control and some typicalalgorithms. The main concern is to provide new algorithms to cure the deficiencies whichexist in current algorithms. To this end, we provide the following three algorithms: therate and queue length based adaptive FARRED algorithm, the pattern recognitionsupporting PR-ControUer algorithm and the rate anti queue length grounded adaptivefuzzy controller algorithm. Tools such as NS-2, GAWK and MATLAB are introduced tothe simulation, analysis and comparison of RED, PI and these three algorithms. It isfound that AQM algorithms which use rate and queue length as their congestionindicators have faster response speed and shorter convergence time. It is also proved thatAQM algorithms designed on the basis of intelligent control theory have better stabilityand robustness and are more applicable to the complex and ever-changing networkenvironment.