Design And Implementation Of Measurement Data Analysis System For Base Station

With expansion of 5G communication,the market demand on 5G base station is increasing day by day,the product technology of base station develops with direction of multi-band,multi-antenna,and multi-RAT.Therefore,production test items and test data of a base station became huge and difficult to be analyzed.However,the R&D department and the production department of communicate company are independent,which make it difficult to share the production data between them.R&D department which has capability on data processing and SW development,gets small part of the production data before product release only.While,after product delivered by R&D department,production department often ignores test data analysis because they usually focus on improving the delivery rate of factory.The production test data stores in database with low utilization.With the facts,to develop a data mining algorithm which could extract the information of production test situation and product performance is helpful to improve product quality of communication company.By analyzing the user scenarios of production test data,the thesis lists the detailed requirements and corresponding design methods.In the view of production department,the most important target is to improve delivery rate of the factory.The production department can analyze the production test data with method of frequency pattern set mining,so as to achieve the purpose of deleting test items,checking the quality of product resource,verifying the stability of test software,monitoring the environment of test stations,and comparing the production quality of different factories.From perspective of R&D,delivering product with high quality and low cost to factory is the main target.Cluster analysis is applied to group each items of production test data,to check product design defects from production data,and to estimate the design risks of similar products,so as to correct the research model and shorten the developing time in project.The thesis developed a data mining algorithm based on MATLAB and developed a data analysis subsystem.This subsystem consists data acquisition module,requirement management module,mining algorithm module and data presentation modules.Two sub-modules,frequency pattern set mining and cluster analysis is including in mining algorithm module.After coding,the production test data of a base station in 2020 is used to verify the algorithms under different conditions.Independent modules bring out good usability,extendability and readability.