基于T-S模糊模型的TCP网络拥塞控制研究

发布时间：2018-06-29 03:38

本文选题：主动队列管理 + T-S模糊模型　；参考：《东北大学》2014年硕士论文

【摘要】：随着网络规模的快速增长,网络用户和应用的不断增加,网络拥塞成为一个亟待解决的问题。基于路由器的主动队列管理机制,与TCP协议的拥塞控制相结合,是解决目前TCP网络拥塞控制问题的一个主要途径。所以,主动队列管理算法的研究是TCP网络领域的一个研究热点。考虑到TCP网络的强非线性和时滞特性,本文采用模糊控制来设计主动队列管理算法,主要的研究内容有：(1)建立了TCP网络速率流体流模型的T-S模糊模型。针对网络拥塞控制系统的非线性和时变特点,建立T-S模糊模型,采用网络拥塞控制系统的多个基于不同稳态工作点的局部线性化模型来逼近非线性网络拥塞控制系统模型。(2)提出了一种基于观测器的主动队列管理算法。基于TCP网络拥塞控制系统的T-S模糊模型,考虑网络系统中的状态往往不完全可测以及系统中存在不确定性和时变性,设计了基于观测器输出的状态反馈控制器。(3)提出了一种基于输出反馈控制的主动队列管理算法。基于TCP网络拥塞控制系统建立的T-S模糊模型,设计静态输出反馈控制器；进一步考虑TCP网络系统中存在的不确定性,给出系统稳定的充分条件,并且利用LMI工具箱求解控制器的参数。(4)提出了基于极小极大控制的主动队列管理算法。考虑TCP网络系统中网络参数的高度变化以及存在的UDP流干扰等因素,在T-S模糊规则中加入干扰,设计极小极大控制器,避免了算法的保守性和放缩性引起的误差；同时考虑系统中的参数不确定性和外界干扰,利用LMI求解极小极大控制器参数。利用MATLAB对以上算法进行仿真,结果表明,上述三种算法均能使路由器缓存队列长度在多种网络仿真环境下快速的收敛到期望值附近,并具有较好的稳定性和鲁棒性。最后对全文作出总结,并提出了下一步研究的方向。
[Abstract]:With the rapid growth of network scale and the increasing of network users and applications, network congestion becomes an urgent problem. The active queue management mechanism based on router, combined with the congestion control of TCP protocol, is a main way to solve the congestion control problem of TCP network. Therefore, the research of active queue management algorithm is a hotspot in the field of TCP network. Considering the strong nonlinearity and time-delay characteristics of TCP network, the active queue management algorithm is designed by fuzzy control. The main research contents are as follows: (1) T-S fuzzy model of TCP network rate fluid flow model is established. Aiming at the nonlinear and time-varying characteristics of network congestion control system, a T-S fuzzy model is established. Several local linearization models of network congestion control systems based on different steady-state operating points are used to approximate the nonlinear network congestion control system models. (2) an observer based active queue management algorithm is proposed. Based on the T-S fuzzy model of TCP network congestion control system, the state of the network system is often not completely measurable, and there are uncertainties and time-varying in the system. A state feedback controller based on observer output is designed. (3) an active queue management algorithm based on output feedback control is proposed. Based on the T-S fuzzy model of TCP network congestion control system, the static output feedback controller is designed, considering the uncertainty of TCP network system, the sufficient condition of system stability is given. The LMI toolbox is used to solve the parameters of the controller. (4) an active queue management algorithm based on minimax control is proposed. Considering the high variation of network parameters and the existence of UDP flow interference in TCP network system, the interference is added to T-S fuzzy rules, and a minimax controller is designed to avoid the error caused by the conservatism and scaling of the algorithm. At the same time, considering the parameter uncertainty and external disturbance in the system, the minimax controller parameters are solved by LMI. The simulation results with MATLAB show that the above three algorithms can make the router cache queue length converge to the expected value quickly in many network simulation environments, and have good stability and robustness. Finally, the paper summarizes the full text and puts forward the direction of the next research.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.06

【相似文献】