基于通信网络日志的故障诊断的研究

发布时间：2018-05-01 16:30

本文选题：网络日志 + 故障诊断　；参考：《北京邮电大学》2014年硕士论文

【摘要】：随着现代信息技术的高速发展,企业和个人对通信网络的安全性和可靠性要求越来越高,因此快速的网络故障诊断和定位显得越来越重要。而利用通信网络运行中产生的日志信息来帮助进行网络诊断,成为近年来网络故障诊断研究的一个热点。本文基于与企业合作方的科研项目,通过对日志信息设置主动监测项,完成信息抽取。对抽取出的信息,从时间维度进行关联规则挖掘,生成频繁告警序列；从空间维度利用故障树分析,定位根源网元故障节点,从而完成网络故障诊断。具体来说,有以下三方面的工作内容：第一,通过研究Apriori和WINEPI算法,改进其在扫描次数和存储结构上的不足,结合日志中告警信息特点,提出了一种基于前缀树结构的Prefix-WINEPI频繁序列挖掘算法。通过对比实验,验证了该算法在执行时间和处理增量问题方面比WINEPI算法更加高效。第二,针对日志信息的字段特征,设置了主动监测项,利用公式计算的方法,快速抽取日志中有用信息,减少无关信息干扰；同时,提出了一种将时间维度的告警关联分析和空间拓扑的故障树分析相结合的诊断方法,应用于网络故障诊断。第三,将理论工作落实到实践。通过书写诊断工具设计文档,完成算法设计和数据存储设计工作；利用Java Web编程和Web前端的相关技术,实现了一个具体的日志诊断工具。解决了故障诊断问题。本文将理论研究与具体的日志诊断工作相结合,验证了上述研究所提供方法的有效性和高效性,较好地解决了科研项目中通信网络日志故障诊断的实际问题。
[Abstract]:With the rapid development of modern information technology, enterprises and individuals demand more and more security and reliability of communication network, so the rapid network fault diagnosis and location are becoming more and more important. And using the log information generated in the communication network to help the network diagnosis has become a research on network fault diagnosis in recent years. A hot spot.
Based on the scientific research projects with the partners of the enterprise, this paper sets the active monitoring items for the log information and completes the information extraction. The extraction information, the association rules mining from the time dimension, the frequent alarm sequence, the fault tree analysis from the spatial dimension, the location of the root source network fault node, thus the network fault diagnosis is completed. In body, there are three aspects of the following work:
First, by studying the Apriori and WINEPI algorithms and improving the shortage of the number of scanning times and storage structure, a Prefix-WINEPI frequent sequence mining algorithm based on the prefix tree structure is proposed in combination with the characteristics of the alarm information in the log. The comparison experiment shows that the algorithm is compared with the WINEPI algorithm in the execution time and the processing increment. More efficient.
Second, in view of the field characteristics of log information, the active monitoring item is set up, and the useful information in log is extracted quickly by the method of formula calculation. At the same time, a diagnosis method which combines the time dimension alarm correlation analysis with the fault tree analysis of the spatial topology is proposed, which is applied to the network fault diagnosis.
Third, carry out the theoretical work to practice. Through the writing diagnosis tool design document, complete the algorithm design and the data storage design work; use the Java Web programming and the Web front-end related technology, realized a specific log diagnosis tool, solved the fault diagnosis problem.
This paper combines the theoretical research with the specific log diagnosis to verify the effectiveness and efficiency of the method provided by the above research institute, and better solve the practical problem of the fault diagnosis of the communication network log in the scientific research project.

【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP311.13;TP393.06

【参考文献】