The Research And Applicationg Of Processing Massive Data In Mobile Network Performance Analysis System

Mobile network performance analysis system is one of the core mobile service support system. Through collecting and analysising the data in mobile communications network at runtime, the mobile network performance analysis system provides a scientific basis for assessing the operation state of the mobile communication network. In addition, through the statistical analysis of historical data, the mobile network performance analysis system can accurately draw the operation of mobile communication network in each time period and found the network operation variation of mobile communication network in various time periods, which providing reliable data assurance to network manager when maintain the mobile network.However, The rapid development of mobile network will bring the explosive growth of business data in mobile network performance analysis system, making data in mobile network performance analysis system with massive and multidimensional characteristics. How to provide efficient processing capability to growing massive and multidimensional data has become a serious challenge for mobile network performance analysis system.An index structure, RAKDB-Tree, which can efficiently deal with massive and multidimensional feature data, is proposed in this paper for the characteristics of mobile network performance analysis system. Based on the idea of the approximation, RAKDB-Tree divides the multidimensional data space into approximations. Learning from KDB-Tree indexing ideas, RAKDB-Tree indexes the divided approximation. According to the characteristics of RAKDB-Tree, the paper proposes query, insert, delete, split, consolidation and restructuring algorithms. With the insertion and deletion of data, these algorithms ensure the RAKDB-Tree can be self-regulating and self-optimizing, and make the tree remain optimum state.Finally, we build a test environment for mobile network performance analysis systems, and select a province Unicom’s actual operational data for testing. Test results validate that RAKDB-Tree can effectively deal with the massive and multidimensional data.