Research On Network Congestion Control Based On Genetic Algorithm

With the rapid development of the Internet, network multimedia services are gettingdiversified day by day. The number of users and the network applications is increasingdramatically. Network congestion is becoming the bottleneck which restricts the networkdevelopment and application. So the congestion control of the Internet is crucial. As the mainway to improve the network system performance and reform the service quality, the research ofnetwork congestion control problem has important theoretical significance and application value.This thesis studies the network congestion control method from the angle of local and globalrespectively. In view of a single node of the network system, active queue management(AQM) isa kind of widely used method and the effect of network congestion control is good. It is aresearch hotspot in the field of internet congestion control. In this paper, on the basis of reviewingthe state of art in AQM algorithm, the intelligent control algorithm is introduced to design thenew AQM algorithm which significantly improves the performance of the queue. Starting fromthe global network system, aimed at the key node and key link in the network, the new routingalgorithm is implemented which balance the network load, reduce the packet loss rate and timedelay, improve the link throughput and obtain a high QoS network environment through thenetwork path optimization. The contributions of this thesis are as follows:Based on the relative changes rate of queue length and the link rate, with a fuzzy neuralnetwork controller whose parameters has been optimized, a method of network congestioncontrol is presented to improve the network performance. Two relative errors are introduced ascongestion notifications, the first is the relative error between current queue length and thedesired value, the second is the relative error between the link rate and link capacity. By using theimproved genetic algorithm for the parameters optimization of fuzzy neural network controller,the network congestion control is implemented. In the conditions of large delay and sudden flow,the stability of the algorithm and the control effect is satisfactory.It studies the technical characteristics and the implementation process of QoS, and analyzesthe advantages and disadvantages of the existing routing algorithm. From the perspective ofrouting optimization, the article propose a routing optimization algorithm of geometric route inmobile Ad hoc Networks(MANETs) based genetic algorithm. The goal is to minimize route delay,balance the distribution of load as far as possible in there are ample free resources on the link tofacilitate accept more requests in future expediently and achieve the purpose to improvethroughput and avoid congestion while meeting the bandwidth, delay, cost and many other QoSmetrics.