| In recent years, with the development of computers and the Internet, the network has become an important part of people's daily work and life. As the rapid increase in Internet users, video, video streaming and other multimedia on a large number of network, the limited bandwidth resources can not meet the growing needs of customers, network congestion has become a serious problem. These problems are mainly manifested in the increase of packets transmission delay and loss probability, the upper drop in application performance. Therefore, congestion control has become an urgent problem to be solved.TCP Vegas is a TCP congestion control mechanism which was proposed by Brakmo.L.S in 1995. It determines the state of the network through the changing of round-trip delay. The biggest difference of TCP Vegas and TCP Reno is the mechanism to judge the network available bandwidth. TCP Reno uses packet loss to detect network congestion situation. In fact, the measures adopted after the occurrence of the congestion. But TCP Vegas judges the network state through the RTT, this is the congestion precautionary measures before the occurrence. Such forward-looking can better predict the use of network bandwidth. It has better fairness, better efficiency, and better adaptability to small cache. The experimental results show that Vegas has a higher throughput than Reno(throughput increased by 37%-71%), smaller packet loss rate, higher stability. Unfortunately, TCP Vegas suffers several problems that inhere in its congestion control mechanism, these include issues of rerouting, persistent congestion, fairness, and network asymmetry. In the high-bandwidth networks, delay-based congestion control mechanism has been proven to get better performance. Therefore, in order to accommodate future network development, research on the Vegas congestion control algorithm is significant.This paper reviews the current status of network congestion, introduces the TCP Vegas congestion control algorithm, proposes research based on background. Then introduce their research results, the research work are as follows:First, in the wired network environment TCP Vegas could not be reasonable to share the bandwidth when it co-exist with TCP New Reno and the poor fairness between old and new connections. In order to solve the problems existing in the application of TCP Vegas, an improved algorithm named TCP Vegas-W was proposed. The algorithm adjusts the congestion window by the formula cwnd=cwnd+(β-Δ)*succ, and adjusts the thresholdαandβto make more data packets in the router's cache. The experimental results show that improved algorithm has a higher throughput than Vegas, has a better fairness between old and new connections.Second, TCP Vegas changes too early from slow-start phase into congestion avoidance phase in the large BDP (bandwidth delay product) links and the fairness bias against connections with higher bandwidth. To solve the problems, we propose a modification to the slow-start phase, which changes the increase manner of congestion window to between exponential growth and linear growth, in congestion avoidance phase through the comparison between the current congestion window and the previous one to determine the status of network.. The simulation results show that the TCP Vegas-b improve bandwidth utilization, maintain a good fairness and friendliness.Finally, give a summary of this article. Propose some new idea that in order to work continue to develop in the future. |
PDF Full Text Request