Mobile Communications Alarm Correlation Rules Mining

Nowadays, the telecom market becomes more and more finely divided. Telecom operators are making efforts to provide services that are different from and superior to competitors to enhance their competitive capabilities. It is more and more important for them to develop an efficient, function-rich analyzing system and make some policies to suit to market and enterprises. In recent years, the Chinese telecom operators gradually adopt data warehouse and data mining system in CRM and Billing system, having achieved much success. However, there is still no such implementation in network management for mobile communications. On the other hand, network management data becomes vital for network upgrade, expansion and maintenance.In year 2004, Network Management department of Shandong mobile telecom Ltd raised ideas to establish "A Data Mining System for Fault Correlation Rules for alarm data in Mobile Communications";and its aim is to improve current comprehensive mobile communications network management system by expanding data source, increasing data volume, and enriching statistical analyzing models. It is expected to avoid prejudices just from so-called expert analysis, to analyze from actual network running situation, so as to promote service assurance quality through maintenance of relevant groups in the big team, evolving from "equipment centric, network centric" to "service centric, customer centric" .A network management data warehouse is established based on the alarm data of mobile communication devices and networks in this study. Meanwhile, it implements the mining function on some subjects so that it can provide the strategies and sustentation for the Network Support Platform (NSP)of Shandong mobile telecom Ltd. In the study, based on fault data of the communication in the mobile network of Shandong province, the Alarm Correlation Rules Mining System (ACRMS) is constructed. The main part of the project includes the total architecture of ACRMS, objective and primary functions, the method of design and exploring tools. Besides, depending on the practical requirement, we build the database which is used to keep the alarm data. This database can support the operation analysis strongly in mobile communication. Using the WINEPI miningalgorithm of correlation rule, ACRMS system can analyze alarm relativity with the concentrated alarm data from database. Synchronously, we apply the alarm correlation rule in real-time alarm supervision. It has become a powerful tool for Shandong mobile telecom Ltd. to analyze and deal with the network data completely in time. This system can provide the sustentation for making decision in order to help staff do general optimization and maintenance.The study and fulfillment of this subject will explore a new way for data warehouse and data mining technologies in mobile communications network management and optimization. This subject has quite innovation, and will provide excellent demonstration for data warehouse in the national mobile telecom network management fields with the huge social and economic benefit.