基于滑模控制的TCP网络主动队列管理算法的研究

发布时间：2018-10-29 21:14

【摘要】：随着互联网的广泛普及,许多新型的网络应用不断出现,网络用户数量的急剧增加以及网络带宽的不足直接导致网络拥塞的发生,大大降低了服务质量(QoS)。由于基于路由器的主动队列管理机制可以有效解决网络拥塞问题,所以对主动队列管理算法的研究成为了TCP网络研究领域的一个热点问题。本文在分析以往算法的基础上,从控制理论的角度分析TCP网络系统,将滑模变结构控制理论融入到主动队列管理算法的设计中,并基于网络模拟软件NS-2 (Network Simulator)对所设计的算法进行了仿真研究,验证了算法的有效性。本文的主要研究工作如下：以TCP网络系统的非线性动态模型为基础,将系统中的参数摄动和外界干扰等非理想因素整合成一个总的不确定,基于反步方法提出了一种滑模变结构网络拥塞控制算法。由于普通滑模控制只在滑动模态运动阶段具有鲁棒性,而在系统的到达阶段对参数摄动和外界干扰等并不具有鲁棒性,所以基于全局滑模控制进行拥塞控制算法的设计,即在滑模面的设计中引入全局滑膜因子,消除了系统的到达阶段,使得系统在全响应过程都具有鲁棒性。基于NS-2的仿真结果表明所设计的算法能够使路由器中的瞬时队列长度稳定在期望队列长度附近,具有良好的稳定性和全局鲁棒性。在TCP网络拥塞控制的研究中,所设计的主动队列管理算法要求具有很好的收敛性能,因此将有限时间滑模控制思想应用于TCP网络主动队列管理算法的设计中,使得系统能够在发生拥塞后的有限时间内恢复到正常状态。为保证系统在全相应过程具有有限时间特点,提出了全局有限时间滑模拥塞控制算法,即在设计滑模面时综合考虑传统滑模控制与有限时间滑模控制的优点,在滑模控制的到达阶段和滑动模态运动阶段都运用有限时滑模控制的方法,从而可以使得系统状态能够快速、精确的收敛到平衡状态。基于NS-2的仿真结果说明所设计的算法能够使系统状态在有限时间内收敛到平衡点,并具有良好的鲁棒性。最后对全文作出总结,并提出了下一步研究的主要方向。
[Abstract]:With the wide popularity of the Internet, many new network applications are emerging, and the sharp increase in the number of network users and the lack of network bandwidth directly lead to the occurrence of network congestion, which greatly reduces the quality of service (QoS). Because the router-based active queue management mechanism can effectively solve the network congestion problem, the research of active queue management algorithm has become a hot issue in the field of TCP network research. Based on the analysis of the previous algorithms, this paper analyzes the TCP network system from the point of view of the control theory, and integrates the sliding mode variable structure control theory into the design of the active queue management algorithm. The algorithm is simulated based on the network simulation software NS-2 (Network Simulator), and the validity of the algorithm is verified. The main work of this paper is as follows: based on the nonlinear dynamic model of TCP network system, the nonideal factors such as parameter perturbation and external disturbance are integrated into a total uncertainty. A sliding mode variable structure network congestion control algorithm based on backstepping method is proposed. Since the general sliding mode control is robust only in the moving phase of the sliding mode, but it is not robust to the parameter perturbation and external disturbance in the arrival stage of the system, the congestion control algorithm is designed based on the global sliding mode control. That is, the global synovial factor is introduced into the design of sliding mode surface, which eliminates the arrival stage of the system and makes the system robust in the whole response process. The simulation results based on NS-2 show that the proposed algorithm can stabilize the instantaneous queue length in routers near the expected queue length and has good stability and global robustness. In the study of congestion control in TCP networks, the proposed active queue management algorithm requires good convergence performance. Therefore, the idea of finite time sliding mode control is applied to the design of active queue management algorithm in TCP networks. The system can return to normal state in a limited time after congestion. In order to ensure that the system has finite time characteristics in the whole process, a global finite time sliding mode congestion control algorithm is proposed, that is, the advantages of traditional sliding mode control and finite time sliding mode control are considered in the design of sliding mode surface. The method of finite time sliding mode control is used in both the arriving stage of sliding mode control and the moving stage of sliding mode, so that the state of the system can converge quickly and accurately to the equilibrium state. The simulation results based on NS-2 show that the proposed algorithm can make the system state converge to the equilibrium point in finite time and has good robustness. Finally, the paper summarizes the full text and puts forward the main 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】：TP393.07

【相似文献】