Application Of Data Mining In CDMA Network Analysis

Since 1980's,the Mobile Communication develops rapidly.the management of mobile communication is facing more and more complex challenges,and the data acquisition,analysis and optimization of the mobile communication network have put forward higher requirements.Therefore,China Telecom launched the development of a CDMA wireless network centralized control platform,To assist the management of CMDA network planning and optimization,improve the network optimization,engineers spend too much time to collect data and analysis by manual handling.finally create a network with the characteristics of our province.In the construction process of the platform,the main work of the author is collecting data from various server data,write database stored procedure and Java background procedures to analysis the data,then use the MVC model to show the output data in the front platform,test and verify the analysis results with the network optimization engineer,production system output documents,etc.The system has achieved the design requirements in general,but the function of some modules in the system also has the problems of the low efficiency of data acquisition,strong subjective factors in the analysis process,and the analysis result is not ideal.The goal of this paper is aiming at the deficiencies of CDMA network management system in the process of data acquisition,data processing and analysis data.use the theory of data mining and data distributed system to rethink these process and improve these process.The main work is:design and write data extraction and preprocessing module and improve the process of data preprocessing and provide pure data for the following algorithm;using clustering method to improve the shortage of projects in the grid cluster to make the clustering result more concentrated;using Hadoop to improve low efficiency of CDT in the project,using Apriori to find frequent set,analysis user needs the recommended results;using BP artificial neural network to realise the platform from passive warning to active warning.