A Study Of The Router Congestion Control Based On Self-similar Model

With the rapid development of Internet,people have an ever-growing demand for the quality of service of computer networks(QoS).Presently,in the high-speed network multimedia's application not only has the very high bandwidth requirement to the network,but also requires intelligence transmission,low delay,the low vibration to be provided the end-to-end control and guarantee of QoS.Congestion control mechanisms are very important in guaranteeing the network quality of service. Moreover,a suitable congestion control mechanism is the basis for other QoS mechanism normal work.And nowadays,congestion control mechanism of Internet and quality of service are the central issues of the current research.However,in recent years,a series of test result show that network traffic flow is self-similar and long-range dependent(LRD),which break the old basic suppose that network traffic flow is short-range dependent(SRD).There are many aspects of the network,such as network analysis,network design,network control,network performance estimation and so on,affected by the self-similary characteristic of network traffic flow.That is,the conventional method has not been applied to these traffics.And the traditional network model ignores the important characteristics of network traffic flow,and fail to reflect the real situation of the network.Firstly,this paper does some research on the congestion control mechanism of Internet.The active queue management mechanism(AQM) which the IETF recomends,is the essential technology based on the router congestion control,which combines with the TCP end-to-end congestion control,being a main method to solve the congestion control question of the present Internet.Secondly,the self-similar,LRD theory and the calculation of Hurst coefficient is introduced in detail.EBP(Embedded Branching Process) measure is conclued,which is advantage to the seven traditional ones,based on experiments and theories.Finally,RED is improved based on the self-similar theory.A queue management algorithm,called ascending semi-Cauchy distribution and Hurst coefficients Adaptive Random Early Detection,CHARED for short,was proposed based on ascending semi-Cauchy Distribution and real-time Hurst coefficients calculated by EBP. CHARED algorithm,which can adjust dynamically the packet dropping probability of the queue management algorithm by forecasting the requriment of traffic flow to the network in the future based on the self-similar and LRD,can make full use of network resources and avoid network congestion effectively.At the same time,the principle of NS2 and the simulation process to generate the self-similar traffic data by NS2 is introduced.The algorithm is verified in NS2 network simulation machine as well.By the indication of a series of simulation experiments,CHARED can validly adapt the change of network flow effectively.Simulated results demonstrate that the CHARED can be used to reduce packet loss ratio,queue delay and queue delay variability.