Research And Application Of Data Mining Applied To Alarm Correlation Analysis In Integrated Network Management System

With the high-speed development of telecommunications industry, telecommunications network is consist of many of private networks which are interconnected,so that network resources and business are becoming more and more closely interlinked.As a result,the development of telecommunication network management tends to be comprehensive and intelligent.Integrated network management system achieves to monitor the sub-network management systems together.However, due to the increasing scale of telecommunication networks and more and more complex construction,the type of alarm is rich and a large amount of alarm is generated at the same time.It is very difficult to find the root-alarm and localize the fault source accurately.In order to solve this problem,alarm correlation analysis needs to be introduced.Alarm correlation rule is necessary for alarm correlation analysis.So it is one of the key points to get the rules to achieve alarm correlation analysis system.Because the alarm rule which is got by network management staff's experience is not good enough to adapt to the frequent change of telecommunication network,alarm data in the network management system needs to be analyzed and processed by other accurate methods in order to obtain alarm rule.Based on research of theories and algorithms about alarm correlation analysis and rule mining,the paper presents the method of improving efficiency of rule mining algorithm,reducing the number of database scanning and I/O load by introducing the technologies of analytic hierarchy process(AHP) and time series.Then,this paper design and implement the rule mining system which uses the improved algorithm and realizes the function of rule self-learning. The process of achieving the system is divided into three phases,including alarm data preprocess,rule mining and post-process of rules.Application of alarm correlation analysis will be perfect in the network management system,if both rule mining system and rule engine system are achieved in the same system.Finally,the system will be used by the SHUOHUANG railway integrated network management system,in order to filter alarm,analyze root-alarm,localize fault source,improve the efficiency of network operation and ensure the network to be normal.