基于拓扑依赖关系和滑动时间窗口的网络事件关联研究  

Research on network events correlation based on topological dependence and sliding window

作　　者：杨家海[1] 王继龙[1] 吴建平[1] 

机构地区：[1]清华大学信息网络工程研究中心,北京100084

出　　处：《微电子学与计算机》2004年第8期43-47,50,共6页Microelectronics & Computer

基　　金：国家"863"计划项目(2003AA103110)

摘　　要：文章提出一个基于拓扑依赖关系和滑动时间窗口的事件关联模型和算法,滑动时间窗口机制用于实现时间域的事件关联,从拓扑结构知识提取的被管对象间的依赖关系用于实现空间域的事件关联。该方法能够动态地自适应网络结构的变化,减少规则学习时间;本方法将较复杂的拓扑依赖关系计算提前到事件关联前一次性进行,因此将事件关联算法的复杂度从O(M2)降低到O(M×N姨)。The paper proposes an events correlation model and algorithm based on topological dependence and sliding time window. Where the sliding time window i s used to correlate events in time domain, and the topological dependence inform ation extracted from network topology discovery is used to correlate events in s pace domain. This approach can dynamically adapt to the change of network struct ure, shorten time cycle of rule learning. As the relatively complex computation of topological dependence is finished prior to the online event correlation, the complexity of the algorithm is reduced from O(M2) to O(M×N).

关 键 词：网络管理 故障管理 事件关联 拓扑依赖关系 管理对象 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]