A Study On Internet Congestion Control Algorithm Based On IPv6 Packet Marking

Internet has experienced an explosively growth in the last more than ten years. Although the current network can provide very high bandwidth, the appearance of varied application, as well as the sharp increase of network users has made network congestion a significant problem. Network congestion often results in the decline of quality of service (QoS) such as transmission delay and throughput, while the network resource utilization like bandwidth and buffers are also affected seriously. How to avoid and control network is necessary for improving the performance and utilization of network.In this paper, we firstly give a review of the progress of the network congestion control research, and the current situation of this field was analyzed. From the view of implementation place, the congestion control strategies include the end to end congestion control implemented at end-point and the queue management implemented at the router. The core of end-to-end congestion control is to make the end-point adapting to network's actual situation through adjusting the sending rate. In common, end-to-end congestion control includes four parts: slow start, congestion control, fast retransmission and fast recovery. Some researchers have made much study and improvement on it, but principle of"add-increase multi-decrease"has not been changed. Along with the technical progress, for high speed network, the end-point can not use the resource according to the actual network situation.In this paper, a new congestion control strategy which includes end-to-end congestion control and queue management was presented. For a data sender, it will adjust its data rate to adapt to the actual network situation if the sender has been informed of the change of network situation. Chapter four introduces the principle and implementation of congestion control algorithm based on bandwidth estimation.About the network situation, we detect it based on queue management. The algorithm can analyze data packages and calculate the number of flows the router has processed. Increase of the number of flow means the payload of the router will be enhanced, at this situation, the router make the sender be informed of this event by package marking. For IP based internet is a connectionless package switching network, it make some trouble for the statistic of the number of flows, chapter five gives a method to solve this problem.In final part of this paper, we introduce the architecture and main components of NS. Then we discuss how to setup simulation scripts and analyze simulation results. Chapter six gives a summarization of this paper and present the next work we will do. Key words: internet congestion control, bandwidth estimation, queue management...