摘要
网络拥塞会严重影响互联网性能,甚至导致网络崩溃。在发生拥塞时,准确检测异常链路,定位拥塞位置、估计拥塞程度,使拥塞在早期就被及时控制,避免拥塞扩散,对网络管理和监控具有重要研究意义。本文根据网络透视技术思想,提出一种基于EM算法的拥塞检测方法,从一个新的角度检测链路拥塞的发生。实验证明该方法有效,对大规模网络管理和监控具有重要意义。
Network congestion greatly threats the erformance of network and even results in network paralysis. It is of the utmost importance for network management to detect and locate the source of network congestion in the beginning of its spreading. In this paper, a new method based on network tomography and Expectation-Maximization algorithm was proposed for congestion link detection. Experiment results indicate the method is effective and of great importance for network management and monitor.
出处
《微计算机信息》
北大核心
2007年第02X期121-123,共3页
Control & Automation
基金
国家"863"计划基金资助项目(2002AA142020)
国家自然基金项目(60203021)
