| In recent years,with the continuous refinement of network technology and the diversified development of its applications,network traffic measurement technology and network congestion control mechanism have been widely focused by researchers and organizations at home and abroad.In the current high-speed network environment,the network traffic measurement work puts forward high requirements on the computing power and storage space of the measurement equipment.The traditional measurement method is difficult to adapt to the traffic measurement in the high-speed network environment due to its own scalability limitations.The network traffic measurement method based on data flow is an important means to measure high-speed network traffic at present.In addition,with the intergenerational replacement of the Internet,it is difficult for the traditional TCP/IP network congestion control scheme to meet the requirements of high bandwidth,low latency and fair network performance at the same time when dealing with massive burst data flows on high-speed network links,and the active queue management algorithm is the mainstream method of network congestion control at present.This paper mainly focuses on network traffic measurement and congestion control.First of all,this paper starts from the overestimation problem of Count-Min Sketch.On the basis of Conservative Update Sketch proposed to improve the accuracy of CMS,and proposes a Flag counting-based sketch structure--CUSF,which aims to set a Flag flag bit to detect whether packets mapped to the same counter belong to the same data stream,and then adopt different counting schemes to improve the counting operation of the counter to improve the measurement accuracy,and finally execute the scheme by simulating multiple scenarios.The experimental results show that compared with the CUS algorithm,the average high valuation of the data stream of CUSF algorithm is reduced by more than15%,which can effectively improve the estimation accuracy of the data flow.Secondly,after obtaining data traffic through network measurement,it is necessary to analyze the network traffic by protocol.In this paper,a network topology model is constructed in the NS-3network emulator,and the performance of various congestion control algorithms is tested.It is found that the interval of adjusting the Co Del algorithm The parameters will have an impact on the performance of Co Del’s average queue length,number of lost packets,and throughput.Previously,the target and interval parameters of the Co Del algorithm were almost regarded as fixed(target=5ms,interval=100ms).For this reason,this paper adopts Several sets of simulation experiments have been carried out,and the experimental results have verified that adjusting the interval parameter does have an impact on the performance of Co Del,such as the average queue length,the number of lost packets,and the throughput.Finally,based on the first two works,this paper proposes a sketch-based adaptive active queue management method,s Co Del,for the problem that the Co Del algorithm cannot identify the network congestion state at a fine-grained level and discriminate packet loss,combined with the CUS structure.s Co Del first performs the network packets in the link for queuing delay,and then uses the sketch structure for packet frequency counting percentage judgment.Through the simulation experiment in NS-3,the results show that the s Co Del algorithm is in While maintaining similar throughput performance,it achieves lower average queue length and higher fairness than the RED,PIE,and Co Del algorithms.Compared with the Co Del algorithm,the average queue length is reduced by about85%,and the fairness index is comparable to the RED algorithm than an increase of about 5%.On the basis of quickly alleviating network congestion,the s Co Del algorithm can implement differentiated packet loss policies,ensure fairness between data streams,maintain network availability and stability,and improve network service quality. |
PDF Full Text Request