The Research Of Network Congestion Based On Control Theory Method

Computer network has grown so fast from last two decades, and made communication much easier than before. However because the increase of date flow, sometimes it turns network into congestion and more seriously. So an effective network control becomes the key factor of network robustness, at the same time, it also takes economic of networking into account, therefore it is really important to study how to control network congestion.Computer network is a complicated giant system. With control theory method, we can effectively deal with network system modeling, forecasting and controlling of network congestion, etc. This essay mainly focuses on the research of network congestion problems based on control theory method. Main field and content are as follow:(1) Analyze the reason why congestion happened and TCP/IP congestion control system. Extend function of network simulation software NS2 to finish simulation of network system modeling, forecasting and controlling of network congestion, and work well.(2) Analyze network system model, from the point of data transferring, take down the network system to individual sub-object according to the situation of network facilities, modify network mathematics analyzing model through setting up sub-object model. It shows the validity and reliability of this method through emulating and analysis.(3) Give a network congestion detection method based on fuzzy forecasting. Use this method to get data size of the next time from sent party according to fuzzy forecasting, combine with the length of present queue in router, forecasting the length of queue of the next time in bottleneck link, controlling network running situation ahead, putting network congestion detection into reality. It shows the validity and reliability of this method through emulation and analysis.(4) Give a fuzzy adaptive PID network congestion control method based on fuzzy forecasting. This method is applying fuzzy inferencing method to adjust PID controller parameter adaptively based on fuzzy forecasting results, and accomplish...