Optimization Algorithm Of RED On Route Congestion Control

Nowadays, one of the major problems encountered in the development of Internet,network congestion is caused by resource scarcity and uneven distribution of traffic, but canalso cause network congestion problems delayed data transmission, thereby increasing thedegree of network congestion, more serious cases will cause the network to crash under theabove phenomenon has aroused great attention of researchers for network congestioncontrol. Currently, the network congestion problem still is inevitable, if only rely on thesource end congestion control mechanism, it will greatly reduce the utilization of networkresources. The most effective solution to this problem is the queue management algorithms,queue management algorithm is divided into passive and active queue managementalgorithm queue management algorithms. Is now widely used active queue managementalgorithm, it is through the router queue length control in the vicinity of an appropriatevalue, which can effectively reduce the delay of the packet, and can improve link utilization.Thus, a very important study of active queue management algorithms.With the rapid development of the Internet, the existing TCP congestion controlmechanism in many ways has been unable to meet people’s needs, so people will beexpanded to the middle of congestion control aspects of the network, there has been a lot ofrouter-based congestion control strategy. Random Early Detection RED (Random EarlyDetection) is a network Engineering Task Force IETF (Internet Engineering Task Force)recommended router-based congestion control algorithms.The main work is as follows:First, the paper describes the RED algorithm parameter settings for TCP maximumwindow problems are analyzed and Misra from the model theory of stochastic differentialequations, mathematical means of this model, a series of processing, and finally get a thenew estimate random early detection algorithm parameters-maximum drop probabilityapproach.Secondly, detailing the RED algorithm re-set the maximum drop probability approach:For an already established network, when setting the parameters, the maximum thresholdis usually set to2to3times the minimum threshold, the expected size of the queue lengthcan be based on the actual needs to be set. If the burst is stronger or more conservative forthe delay, it was slightly smaller this value; If the queue length packet loss will be moreconservative set too larger more. Thus, the maximum drop probability estimation methodproposed in this paper really only a minimum threshold and the queue length, and when thefuture of these two parameters determined by a formula can be derived to calculate the maximum drop probability values.Finally, NS-2(Network Simulation Version2) simulation, the value of re-using themaximum drop probability of throughput and delay settings RED routing and routingDropTail were compared according to the same conditions and limitations of two differentmaximum window limit the maximum window, the instantaneous queue length under thetwo algorithms, the average queue length and packet loss rate were compared and analyzed.After the final validation of the re-set the maximum drop probability, RED algorithminstantaneous captain, captain average, packet loss rate, source and number of technicalindicators have been optimized connections, so as to achieve a reduced network congestion,improve transmission quality.