Research Of TCP/IP Congestion Control Based On The Method Of Optimization Flow Control

Internet has been developed rapidly in recent years. The inflation of user's quantity and the constant producing of new applications make the crowded problem of the network more noticeable. Because congestion control is a key factor of guaranteeing the robustness of Internet and a foundation of various kinds of management controlling mechanisms and applications, the problem of congestion control about Internet has been a focus being studied of the network all the time.The introduction of this paper introduces the basic conception of congestion, the reason leading to the congestion and the classification of congestion control. It probes into the congestion control mechanism of TCP and analyses the congestion control problems at present.Chapter 2 introduces two kinds of flow controlling mechanisms at first. One is end-to-end flow control mechanism. The other kind is based on flow control running on the IP layer. Then it analyses the problem existing in the two kinds of flow control mechanisms when dealing with the congested phenomenon. At last it recommends in detail the basic principle of the optimization flow control mechanism, the evolution, the advantage and disadvantage while dealing with the congested problem.Chapter 3 puts forward a new scheme-OCC that is a TCP/IP congestion control mechanism on the basis of optimization flow control and proposes its overall design object and algorithm targets. This scheme consists of two parts. In algorithm of IP layer length of queue and the optimization theory are used to estimate the congested degree of the network and to respond to the congested degree through marking packets. In algorithm of delivering end we estimates congested degree of the link and adjusts the speed of delivering according to the number of packets that has been marked. In the scheme of OCC with the cooperation of the IP layer algorithm and the delivering end algorithm, it considers to the speed of delivering and the available width of bottleneck. Also OCC makes all kinds of data flows coexisted friendly and optimizes the performance of the network area involved to it.In order to verify the rationality and validity of the algorithm, in chapter 4 a large amount of simulation experiments have been made to appraise the performance of algorithms. In order to make the experiments environment close to the real network, in the experiments of verifying fairness of algorithm we use self-similar flow as the background flows. The experiments show that the algorithm can make quick response to the change of congested state of the network. Both the fairness within and between protocols and the robustness to the topological structure of the network are very good. The experiment platform is a NS simulator. NS is a kind of network simulation software widely used in the world at present. It has played an important role in analyzing and appraising the performance of the network. All the experiments of this paper are analyzed and simulated through NS.At first chapter 5 explains the necessity of using self-similar flow as the background flow in network simulations and introduces several kinds of commonly used definitions of self-similar. Then it introduces the method of producing self-similar flow in simulation experiments and theway of using NS to produce self-similar flow. Last we discuss the possibility of using the parameter of self-similar in the congestion control.