The Research About Active Queue Management Algorithm Based On Link-load Adaptive

With the development of computer network technology, the scale of the Internet has exhibitedexplosive growth, the number of Internet users has increased fast and IP business has developedrapidly, and these lead to diverse demands. The problem which is caused by the rapid increase ofnetwork traffic leading to network congestion is becoming more and more serious. Besides, thereare higher and higher requirements to the service quality of computer network. In order to improvethe service quality of computer network, congestion control mechanism based on networkintermediate nodes becomes a research focus. Active Queue Management (AQM) is an importantmeans to achieve congestion control based on intermediate nodes. AQM algorithms came into beingone after another.This thesis analyzes deeply several typical AQM algorithms and summarizes the advantages anddisadvantages of each. In order to achieve stable queue length and improve link utilization, thisthesis proposes an improved algorithm named link load adaptive-RED (TTL-ARED). Thetraditional ARED algorithm detects the average queue length to determine the network congestiondegree and adjusts the max drop rate to stabilize the queue length. But this algorithm has weaknesslike network hysteresis, and it cannot adapt to unexpected network traffic. TTL-ARED can measurethe degree of network congestion according to link load changes to stabilize the queue length andguarantee high link utilization. It did certain improvements to the packet drop probability functionto make it better adapt to network congestion control, reduce the rate of packet drop and increaselink throughput. Simulation results show that TTL-ARED can better adapt to different networkenvironments, effectively control network congestion, enhance the stability of the queue length andensure high link throughput by using NS2to value the performance of this algorithm. AndTTL-ARED has better stability and robustness.