|
As the network continues to expand, network management research to become the IT industry is one of the hottest research direction, communication operators need to build a comprehensive, centralized, intelligent network management system for effective management of communication networks, providing customers with high-quality services, improve the competitiveness of enterprises. Communications network integrated network management need to collect and process data from switching, transmission, data, and other professional dynamic environment of massive network alarm data, in order to ensure effectively prevent alarm storms, to ensure attention to the alarm, quickly find, locate and deal with failure, we must design and implement effective alarm filtering mechanism to achieve fast and accurate filtering instantaneous break alarms, invalid alarms and minor alarm, which is building the core communications network integrated network management tasks. This article first alarm data communication network characteristics are analyzed in detail, and elaborates common alarm correlation analysis methods, and association rule mining algorithms. Secondly, when comparing the existing alarm correlation analysis method, discussed how to use data mining techniques to obtain the alarm association rules, and on this basis, build alarm filtering rules knowledge base, as well as the alarm filtering template update knowledge representation and knowledge acquisition method. Finally, based on the results of the analysis presented hierarchical alarm filtering model and template-based alarm filtering program, elaborated knowledge base rule matching and conflict resolution methods, alarm filtering mechanism and its practical application to achieve specific results. Comprehensive domestic network for local communication network research and development work in recent years has it started, but will be applied to data mining and knowledge base technology research alarm filtering mechanism is shorter, so the contents and achievements of research projects have strong theoretical, technical and application value. The main innovation is reflected in the topics: ◇ alarm data based on the characteristics and alarm filtering requirements analysis, definition of a hierarchical structure of alarm filtering model. ◇ design knowledge base built alarm correlation, realized intelligent alarm filtering template generation. ◇ By analyzing classical Apriori algorithm combined with time windows, access to the alarm time correlation rules and applied in practical systems. ◇ using a template-based composite filter technology to effectively achieve the convergence of alarm data layer by layer, alarm filtering flexible configuration and implementation.
|