Research Of TD-SCDMA Network Alarm System Based On Attribute Similarity

TD-SCDMA is a third generation mobile communications standards which with theindependent intellectual property rights. Using time division duplex mode (TDD) makes it hasefficient frequency spectrum utilization and system capacity. This paper is part of the projectof TD-SCDMA network performance prediction analysis and optimization system". And themain study is alarm information management and correlation analysis reported byTD-SCDMA network equipment, which diagnoses network fault intelligent. With the help ofalarm management system, network manager’s workload could reduce effective.With the growing number of users of the TD-SCDMA network, the size of the network isgradually expanding, more and more network devices are in use. In order to ensure thestability of the network and to foresee the failure of network elements within the networkdevice, network manager may be able to analysis to find the fault and its processing of thealarm information timely from the network equipment at all times. TD-SCDMA network isinterconnect by a lot of equipment and links. Network components or equipment failure maycause multiple network equipment and link which associate with it failure, and a large numberof derivatives of the alarm information raised. Therefore, in order to ensure real-time,accurate detection and location network element failure, we must better alarm correlationanalysis methods and mechanisms, quickly and accurately compression derivative alarminformation. Research on and analyze the characteristics of the alarm information. Based onthis proposed, acquisition alarm filtering, alarm normalized, alarm correlation analysis, thealarm information processes, which the system is divided into different functional modules.The same time, the alarm correlation analysis method, and raised the alarm attribute similarityrules-based alarm correlation analysis. At present, the commonly used method of alarmcorrelation analysis are the following methods: based on neural networks, fuzzy logic-basedmethods, rule-based approach, based on data mining association. Alarm correlation analysismethods based on attribute similarity are, strictly speaking, belong to the alarm aggregationassociated quantitative methods, which could be used for the real-time alarm analysis. Bycomputing the similarity of the alarm to determine whether the two alarm informationsecurity incidents for the same alarm. In network fault diagnosis process, we believe that thealarm message caused by the one fault has a very high similarity. These alarms may beprovided by different element, but both detected the failure of the network, or reported by thesame network, but different types or different time. By computing the similarity of the alarminformation to determine whether the alarm information collected alarm information is combined into a richer content. Effectively reduce the workload of manager.Finally，based on B/S architecture, realized the alarm management system. And use thealarm data which actually collected in the Mobile communication network test the system, theresults show that the system for TD-SCDMA network equipment alarm has effective analysisfor the alarm correlation.