Font Size: a A A

The Research Of Improving Adaptive RED Algorithm

Posted on:2012-08-27Degree:MasterType:Thesis
Country:ChinaCandidate:C J ShuFull Text:PDF
GTID:2178330332978588Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
With the rapid development of Internet, new applications are constantly emerging and the number of users is sharply increasing. Internet traffic is consequently explosively growing, which makes the problem of network congestion become more and more serious. Network congestion will lead to a lot of problems, including a sharp decline in throughput, massive packet loss, delay increase and etc., which bring a bad influence on network performance. The research on congestion control has great significance for solving the congestion problem to improve network QoS(Quality of Service).Currently, Internet congestion control primarily relies on TCP window-based end-to-end congestion control and queue management mechanism in routers, which guarantee the stability of Internet. IETF (Internet Engineering Task Force) proposed that routers should use AQM (Active Queue Management) mechanism to control network congestion. AQM has always been a hot issue in IP network congestion control research and a great number of domestic and foreign researchers have been conducting researches on it. Until now, many AQM algorithms have been proposed.In this paper, several classical AQM algorithms are analyzed with their respective advantages and disadvantages compared. Since the traditional ARED algorithm lacks the capability of adapting the target queue length to network load, an improved ARED_IPV (ARED Improved) algorithm is proposed in this paper. ARED_IPV is able to sense network load changes based on average queue length and adaptively adjust target queue length so as to make itself better adapt to different network load conditions, eventually achieving a better tradeoff between delay and throughput. In this paper, instead of using network simulation, we use Click Router nodes to construct network environment and then conduct experiments with real network data to verify the performance of the proposed algorithm. The experiment results show that ARED_IPV achieves the expected improvement to ARED. At last, the application of ARED_IPV in Reconfigurable Network Simulation Platform is introduced. ARED_IPV can supply an efficient virtual network queue management mechanism to Reconfigurable Network Simulation Platform.
Keywords/Search Tags:Congestion control, AQM, ARED, target queue length, Click Router
PDF Full Text Request
Related items