Research On TCP Network Congestion Control Based On T-S Fuzzy Model

With the rapid growth of network, and the increasing number of users and application on the network, the network congestion has become a very important problem. A major way to resolve the problem of TCP network congestion control is the combination of the active queue management mechanism based on the router and the TCP congestion control. Therefore, the study of active queue management algorithm is a hot issue in the field of the TCP network. Considering the strong nonlinear and time-delay characteristics of TCP network, this thesis proposed a fuzzy control method to design active queue management algorithm. The main research contents are as follows:(1) T-S fuzzy model is established for the fluid flow model of TCP network system. Considering the nonlinear and time-varying characteristics of network congestion control system, the T-S fuzzy model is established, which uses different static working point of local linearization model to approximate the model of the original nonlinear network congestion control system.(2) A kind of active queue management algorithm based on observer is proposed. Based on the T-S fuzzy model of the TCP network congestion control system established, considering the situation of the state of the network often can not be fully measured, and the uncertainties and time-varying are existing in the system, state observer is designed for uncertain systems with time delay, and the state of the feedback controller based on the output of the observer is designed.(3) A kind of active queue management algorithm based on output feedback control is proposed. The output feedback controller is designed based on the T-S fuzzy model of the TCP network congestion control system, further consideration of the uncertainty of TCP network system, and gives the sufficient conditions for the stability of the system, and using LMI toolbox to solve the parameters of the controller.(4) An Active queue management algorithm is proposed based on Minimax control. Considering the height change of the parameters and the UDP flow interference in the TCP network system, interference is added in T-S fuzzy rules, and the minimax controller is designed, which avoid the error of the conservatism and shrinkage, at the same time, considering parameter uncertainty and external disturbance in the system, we can use the LMI toolbox to resolve the parameters of the minimax controller.The above algorithm are simulated by the software of MATLAB, the results of the simulation show that under a variety of network simulation environment, the above three algorithms can make queue length of the router quickly converge to the vicinity of expectations, and have good stability and robustness.Finally, the full text is concluded, and proposes the direction of further research.