摘要
现有的网络事件关联系统主要存在以下不足:各被管设备感知的大量告警事件全部传送到管理端处理,会带来很多传输与事件管理问题;现有的网络事件关联方法很不成熟,一般只涉及底层协议告警事件的关联.在分析网络告警与故障关系的基础上,提出了一种在设备这一级驻留代理,采用贝叶斯网络推理技术完成包括应用层告警事件在内的本地告警事件纵向关联方法.并在描述协议栈各层协议实体模型的基础上,给出了利用AdventNet API和ebayes进行系统具体实现的方法.
Current network event correlation systems have several shortcomings such as.. all alarm events emitted by management devices are transferred to the manager, thus many transmission and management problems may happen; existing vertical network event correlation methods are immature as they only concern the low protocol layer alarm events generally. Based on analyzing the relationship of network alarms and faults, a vertical correlation method utilizing Bayesian network reasoning technique for dealing with the local alarm event including application layer alarm events is proposed. At last, the concrete realization method using AdventNet API and embedding ebayes is given after descripting models of protocol layer entities.
出处
《海军工程大学学报》
CAS
北大核心
2005年第6期14-20,共7页
Journal of Naval University of Engineering
基金
国家自然科学基金资助项目(60303023)
关键词
纵向事件关联
不确定性推理
贝叶斯网络
健康函数
vertical event correlation
non-deterministic reasoning
Bayesian network
health function