Research And Application Of Correlation Analysis On Multi-source Monitoring System’s Alarm Data

With the continuous development of the Internet and information technology, networks become increasingly easy to use. As the network has provided us with more and more functions, its own structure is also becoming increasingly complex. The telecommunications network is now becoming more and more difficult to be managed. mance. An equipment’s problem will often affect many peripheral devices, and at this time, each network element depends on the device will send alarms, and if the device or system problems seriously affect the normal operation of other network elements, the problem can spread quickly in the network. When network managers are faced with the so-called "alarm storm", the general manual analysis has been almost impossible to solve the problem. At this point you need to find a way to search for some associations behind these massive alarm data which can allow us to quickly find out the source of the fault. It is called the alarm correlation analysis.The goal of this paper is researching how to dig out useful information in alarm history data which can help to find out the root course of alarms. The first two chapters mainly talk about some basic concepts of alarm analysis and theoretical methods and techniques of dealing with massive alarm data. The third chapter, based on researching two classical algorithms Apriori and FP-growth, talks about the advantages and disadvantages of them and the way of improvement.The fourth chapter gives an improved Apriori algorithm which focuses on solving two problems of the classical one. For FP-growth, this chapter also gives an improved implementation of it. Finally, some experiments of these algorithms have been made.The last chapter introduces a monitoring system which has many advantages. This system is made for collecting alarm data. The structures of the hardware and software have been told in this chapter and some detailed test cases are provided.