Research On Application Of Time Series Analysis Technology In Network Monitoring

With the development of network technology, information exchange and resourcesharing are more convenient. For teachers’ and students’ convenience, campus networkbandwidth expanded year by year, but the utilization is not high. Games, video chat, instantmessaging and P2P take up most of bandwidth, and do harm of the normal work and study.Therefore, how to control the network running, analyze network application or monitor net-work status, is good for improving the effective utilization rate of network bandwidth.The purpose of this paper is to build the prediction model and predict future networktraffic trends by using historical data, through studying the characteristics of campus networkdaily traffic. The application layer protocol analysis technology is used to analyze and statisticthe different kinds of network application. Administrator, according to the result of forecastand analysis, could make network management solution to limite the P2P applications.The main research content of this article is that based on the analysis of the campusnetwork traffic self-similarity, beside of the concept, methods and steps of time series, furtherresearch what mainly focused on the time series analysis of ARMA model, SARMA modeland GARCH model has been done, the anlaysis technologies of the network traffic have beenstudied, including the application layer protocol analysis technology, MFSAM and SNMP.Finally, by comparison with experimental results, a network traffic prediction model is putforward a model based on SARMA to modeling of network traffic history data, and by usingGARCH model to modify the residual.In the paper, with time series analysis technology, the network traffic prediction model isestablished to simulate and forecast the future change of the flow.Then Port, DPI and DFI areall used in the application layer protocol analysis, adopting MFSAM in the process ofprotocol analysis feature matching, which improve the matching speed and accuracy. At lastnetwork equipment is set to achieve to control an IP or application protocol, using SNMP.Combining the above techniques, the network traffic prediction and monitoring systemhas been designed and implemented. The system base on B/S architecture, the front pageusing ExtJS4.0, SSH framework for the MVC pattern development, using JSON for datatransmission, and the system present style of the Desktop. In the last, the design of this systemis introduced, the realization method of each function module has carried on the detailedelaboration, and the system deployment and performance analysis are described.