Alarm Correlation Analysis Based On Association Rules And Its Application In Network Management System

With the rapid development of communication network, network fault management is becoming more and more complex. As an important step, alarm correlation analysis is an alarm information handling technique. By it clients can forecast the operation status of the network and find out the causes making the network run abnormally.There are many methods in alarm correlation analysis. In these methods, the alarm analysis one based on rules can be comprehended easily and consistent with men's logic, so it's applied in the communication network widely. But in the traditional alarm correlation analysis method based on rules, to get the rules depends on the network engineers' knowledge and experiences in great degree. This not only enlarges the working strength but weakens the rules' reliability. In order to solve this bottleneck problem, considering the advantages and confluence of data mining technology and correlation rule, the thesis presents an alarm correlation analysis method which combines data mining technology with correlation rules.This thesis takes the fundamental research as the foundation, combining with the implementation of alarm correlation in Taizhou Telecom data network management system, expounds the application of correlation rule in alarm correlation analysis. Then, to get the correlation rule model, the thesis presents two methods: one is customizing the correlation rules; another one is applying the association rule mining method. At last, taking the alarm data of network management system as the data resource, the thesis mines some association rules and applies them in the alarm correlation analysis system. It's turned out that the association rule mining is feasible.