Research And Implementation Of Alarm Association Rules In Intelligent Network Management System

Along with the rapid development of the Internet, the requirements of thecommunication network infrastructure support is becoming higher, faster, morestable. In the network evolution towards all-IP background, in order to realize highbearing of transmission network for data service, PTN (Packet Transport Network)has become the current package transport network architecture and specifictechnologies. PTN universal application in the communication network, whichgreatly improves the quality of communication networks, is making communicationnetworks more intelligent and convenient.Network devices generally have the ability to perceive abnormal situation. Whennetwork device finds itself or the network is in abnormal status, always notifiesmanagement platform by sending alarm messages. With the increasing complexity ofnetwork scale and structure, centralized network fault management model has beenunable to meet the demand management applications. This brings networkmanagement platform event correlation analysis of the great challenge, its associatedthe accuracy of the result is not certain.This thesis firstly describes the development status of alarm analysis systems incommunication network, analyzes the used methods and the resultsobtained.Secondly introduces the research content and purpose of this study, anddoes some researches in PTN network environment from following aspects: thedefinition,alarm source classification, characteristics, then summarizes alarmmanagement principles and alarm correlation analysis methods, introduces theexisted analytical methods and relevant algorithms. Thirdly analyzes mining stepsfrom these classic FP-growth algorithms, does some optimization in computing headtable which corresponds to conditional FP-tree, and reduces the number of scanningthe FP-tree to accelerate the computation efficiency. Then optimizes the algorithmand introduces the optimized algorithm in accordance with the improvement points,process, pseudo-code, some classic cases, does some tests.Finally, combines with the project PTN Intelligent Network ManagementSystem, does some researches on the system framework, key technologies and design process of alarm module, and analyzes alarm dependent requirements, especially thealarm correlation requirements. Then designs the methods of extracting the alarmtransaction to prepare for the alarm association rules analysis and generatestransaction data sets according to the original data set, uses FP-growth algorithm inorder to improve the mining process, and does the results analysis.