Research Of Congestion Control In Network

Internet has been developed into the basic of today's social life. People are expecting more and more capabilities such as performance, content and safety from network. However, the main kind of service is still "Best Effort" in the Internet. All packets are treated equally. There lacks of efficient queue management. Congestion happens here and there between whiles, which causes the dropping of packets or the jitteriness of transport, and even pervades to collapse the network.There are many factors to influence network QoS. An important one is network congestion. As a crucial method to avoid and control network congestion, congestion control is not only the critical factor of network performance and efficiency improvement, but also the premise of improving network QoS.Combining with the evolution of congestion control mechanisms, the existed technologies for congestion avoidance and recovery in IP networks are reviewed. The unsolved problems and hot spots in subdiscipline, such as TCP flow control, end- to-end flow control and enhanced mechanism in intermediate nodes, are summarized, and the advantages and disadvantages of all kinds of schemes for active queue management are emphatically analyzed. Also the paper presents and discusses the drawbacks of heuristic designing combining with special simulation experiments used by most of network algorithms. Subsequently, a new approach to design and evaluate congestion control algorithm is presented based on control theory, and several interesting directions for future research are provided.The prevail model of congestion control research is that firstly propose a scheme based on experience and intuition, and then verify and improve it through simulations and experiments, and main attention is paid on the functions of the end systems. We analysis the congestion control system as a close loop system and take the router as the feedback controller. So the role of queue management is emphasized and new active queue management algorithms are designed. We discuss structure of the end-to- end congestion control schemes with active queue management (AQM) in IP network fromthe control theoretic point of view. And propose a fuzzy logic based AQM which can improve the robustness of the control algorithms. Give out the detailed simulations.