Measurement And Analysis And Forecast Of Network Traffic Behavior

To solve problems in the management of the campus LAN, this thesis integrates the theory and technique with actual analysis. And a series of research in the characteristic of network traffic behavior of the campus LAN have been done, which offer some supports of making decision in the management of the campus LAN.Firstly, this thesis introduces the basic concept and correlative theories of network traffic and network behavior subject. Secondly, the thesis presents several important components of the research in traffic behavior, which includes network measurement and traffic analysis and traffic forecast. The techniques adopted in this thesis are emphasized such as of passive measurement, sampling and assured time series. Then, the thesis describes the whole process of analyzing the traffic behavior through the example of the campus LAN traffic in this project, which includes designing and realizing an experimental system, gathering traffic data, analyzing traffic behavior, and forecasting.In sampling, possion sampling technology recommended by RFC2330 is used, and the sampling effect is fine. In analyzing, we summarize four characteristics of traffic bursting and analyze these characteristics in point of behavior subject view, which includes smooth changing, single bursting, continuous bursting and interval bursting, In forecasting, we adjust the season change model of assured time series, and put it into forecasting the change trend of traffic today. After comparing forecasting result with actual sampling traffic, our forecast is proved reliable.