【作者】 侯萍；

【导师】 王执铨；

【作者基本信息】 南京理工大学 ， 控制科学与工程， 2011， 博士

【摘要】 随着网络通讯的发展和用户数量的膨胀,网络的拥塞问题也越来越严重。拥塞导致的直接后果是整个网络的性能下降：包括分组丢失率增加、端到端延迟增大、网络吞吐量下降,甚至有可能使整个系统发生崩溃。所以有效地解决拥塞问题是改善网络系统性能,提高网络通讯服务质量的主要手段。设计简单而有效的拥塞控制算法成为网络管理中亟待解决的问题。网络拥塞控制可以看作是一个反馈控制系统,故从控制理论的角度研究网络拥塞控制可以得到更有效的结果。利用控制理论分析现有拥塞控制系统的稳定性,并设计新的拥塞控制算法,具有重要的理论意义和应用价值。因此进一步探索基于控制理论的网络拥塞控制方法是非常必要的。本文研究基于控制理论的网络拥塞控制算法,取得的主要研究成果与创新点如下：1.针对改进的网络简化模型,提出一个大时滞网络拥塞控制算法称为改进的混沌优化PID控制算法(ICPID)。改进的网络简化模型考虑了时滞的影响,利用这个模型,AQM路由器应用改进的混沌优化策略优化PID控制器参数。然后针对PID控制器不能随着变化的网络环境在线调节参数,提出了一种基于增益自适应Smith预估控制和模糊控制的大时滞网络的自适应PID主动队列管理(GAS-FPID)算法。引入增益自适应Smith预估控制器实现滞后补偿,模糊控制器来实现PID参数动态网络环境的在线调整。2.针对网络参数的时变性,将TCP模型描述为具有状态延时和输入延时的系统,设计了基于观测器的状态反馈控制器,利用线性矩阵不等式(Linear Matrix Inequality, LMI)和Lyapunov-Krasovskii理论得到使得网络系统不依赖于延迟的渐近稳定的控制器参数；针对网络参数的不确定性和链路带宽的时变性设计一种主动队列管理算法,将可获得的链路带宽作为标称值,而不可获得的未知的时变链路带宽作为干扰信号,以状态空间的形式描述TCP/AQM模型,用时间域H∞控制方法来解决网络拥塞问题。3.针对具有通信时延的无线传感器网络的拥塞问题,利用图论对无线传感器网络进行建模,借鉴领导者-跟随者的思想设计了一致拥塞控制算法(Congestion Control Based Consensus, CCBC),根据汇聚节点的负载状况,合理的调节所有传感器节点的数据发送速率,应用Lyapunov函数证明算法在变拓扑网络结构下的有效性。4.针对一个用于无线网络拥塞控制算法设计的具有通信时延的流体流模型进行Hopf分岔分析,以通信时延作为分岔参数,证明此模型Hopf分岔的存在性,并应用中心流形和规范型理论推导出确定Hopf分岔方向和分岔周期解稳定性的计算公式,数值仿真验证了结论的有效性。最后,在总结全文工作的基础上,给出了本文后续需进一步探讨的一些问题。更多还原

【Abstract】 With the development of network communications and the expansion of the number of users, network congestion problems are becoming more and more serious. Congestion is a direct result of the performance of the entire network degradation. For example, the packet loss rate increasement, end to end delay increasement, network throughput decreasement and may even crash the whole system. Therefore, the network congestion control is the main way to improve the network performance and reform the quality of sevice.The design of simple and effective congestion control algorithms in network management are the problems to be solved. Network congestion can be seen as a feedback control system. From the perspective of control theory, much more effective results about research network congestion control can be achieved. Using control theory analysis of the existing congestion control system stability and the design of new congestion control algorithms show important theoretical significance and application value. Therefore, it is very necessary to explore the network congestion control method based on control theory. This paper studys on network congestion control algorithms based on control theory, to obtain the main research results and innovation are as follows:1. A new congestion control algorithm called PID control algorithm with improved chaos optimization based on improved model is proposed in the large delay network situations. According to the model, the PID parameters are tuned with chaos optimization in AQM routers. PID controller can not turn parameters online in dynamic network condition. A novel active queue management algorithm for large delay network based on Fuzzy PID control and Gain Adaptive Smith (GAS-FPID) is proposed, which can achieve PID parameters on-line self-adapting by fuzzy control under the dynamic delay network circumstances. And, gain adaptive Smith is successfully introduced into feedback data’s advanced prediction to compensate delay.2. The model of TCP/AQM including the state and the input delay is presented in state variables. State feedback control based observer is introduced to estimate online output of congestion control for AQM router for variable network parameters. According to the Linear Matrix Inequality(LMI) technique and the Lyapunov-Krasovskii theorem, control laws and delay-independent stability criteria for the AQM controllers are derived. An AQM algorithm is presented for network congestion, which is based on uncertain parameters and variable link bandwidth. The available link bandwidth is modeled as a nominal constant value, which is known to the link, plus a time-variant disturbance, which is unknown. The model of TCP/AQM including the state and the input delay was presented in state variables. Then, the network congestion problem is solved by using the time-domain H∞control approach.3. The congestion control algorithm based on Concensus is designed for network congestion over wireless sensor network by distributed dynamic system. The congestion problem is modeled by graph theory, it can be proved that the send rate for all nodes converges to the minimal available bandwidth by the proposed CCBC. Via Lyapunov function, the validity of the proposed algorithm is shown under the varying network topology and time-delay.4. Study on the Hopf bifurcation analysis of a fluid-flow model with time-delay for the congestion control algorithm in the wireless networks. By choosing the communication delay as a bifurcation parameter, the model exhibits of Hopf bifurcation are proved. The formulas for determining the direction of the Hopf bifurcation and the stability of bifurcating periodic solutions are obtained by applying the center manifold theorem and the normal form theory. Finally, a numerical simulation is presented to verify the theoretical results.Finally, based on summary of full text, some problems which need to be further researched are discussed.更多还原