Network Congestion Control Model Research Based On Fuzzy Control

At the end of the twentieth century, with the development of the science and technology,the scale of the Internet is growing exponentially, the number of users that access networkalso growth, a wide variety of network applications has also come to the force, Internet issuffering more and more pressure and problems increasing as well. Network congestion isthat, as one of the important issues in the development of the Internet. Since the problem,some experts and scholars to its research without interruption, the congestion control’sresearch has been an active issue in the field of computer network research.There are many reasons that cause network congestion: a lot of storage space in thenetwork is occupied, the available bandwidth in the link can’t meet the current needs, networkcongestion control mechanisms that has been original designed didn’t take into accountfactors, and some users sabotage. If for some reason later when the network is congested, theoverall link performance will be watered down, the packets sent by the sender is not timely ornot arrive at the receiving end, when the sender receives the receiving end of the ACKresponse will exceed the timeout retransmission time, more serious congestion paralyzed theentire network will not be used. In the current Internet network, we are not able to avoid theoccurrence of congestion, it is necessary to use certain methods and means to avoid andcontrol them. Therefore, congestion control research has important theoretical significanceand application value.In this paper, firstly, the research status at home and abroad were introduced, differentalgorithms were introduced from the source of congestion control mechanisms and the client,some of the system model are analyzed from the control concept point; based on the TCPcongestion control algorithm and the increase integrand decrease periodic window variation,analysis and design of TCP connection throughput model and influencing factors; analysisfairness of TCP congestion control problem from three aspects. Secondly, analysis the source algorithm, and give an improved method. Briefintroduction the history of the theory of fuzzy sets, given the definition of fuzzy sets, as wellas the selection of membership functions and fuzzy rules. The source end of the algorithmdoesn’t take into account the impact of the end of the network parameters on the controlperformance, considering the round-trip delay and router queue length algorithm, the fuzzycontrol applications to the article, to build a two-layer structure fuzzy controller, the resultsshow that both the throughput and window size has increased, and the ideal control effectshowed coexist with TCP Reno.Finally, researched the active queue management algorithm. Introduced the dynamicmodel of the network and Smith predictive control principle, combined the advantages of theboth(fuzzy control doesn’t require accurate model and Smith compensation delay),and thesame time using the parameters of self-correction technology, design a new active queuemanagement algorithm. Experimental results show that the algorithm makes the queue lengthtends to be a certain degree of stability, reduce the impact of the delay jitter queue length,effectively avoid network congestion and improve the quality of service of the network.