Evaluation And Analysis Of LTE Networks Based On Large Data

LTE network assessment is an important phase of network optimization.Data resources of LTE network are used to analyze the actual operation of the network.The evaluation results are used to understand the structure,performance and operating status of the network.After that,the network can be accurately optimized.Network assessment is mainly conducted by using massive Measurement Report(MR)that is generated by mobile terminals in LTE networks.MR data is used to analyze network coverage and interference and provides suggestions for network optimization.The contents of this research include three aspects.(1)MR data is parsed and coverage and interference are analyzed in the network.The massive MR data is analyzed by using Spark and the analytical result is stored in a distributed manner in HBase.Performance optimization problems of analysis and storage in big data are done,and coverage and interference are computed by using Spark.(2)Cell status is detected by means of clustering MR data.The key features that represent cell quality in MR data are extracted through the way of feature analysis,and a stable distributed K-Means clustering algorithm is designed to perform clustering operations.The network status is analyzed based on the data distribution of network level,then the actual status of the cell is computed by using analysis technology of cell level.(3)Network scenarios are identified and analyzed by using multi-dimensional data.Multi-dimensional cell's clustering is achieved by making use of the cell's geographic scenes,KPIs,configuration parameters,then the scenes that represent network key features are analyzed though making use of analysis techniques of the decision tree.Through the above steps,the scenes are identified.The result provides the reference for cell's parameter recommendation.Based on the above-mentioned studies,a spark-based LTE network evaluation software system is designed and implemented,and the performance and scalability analysis of the system is given.The system is developed by using Scala language and C#language,and takes Spark,HBase,HDFS and other components to process the massive data.And this system uses MapXtreme to complete the geographic information processing under the.net platform.The method and system are tested and verified by using the massive data in the actual network environment.The results show that,the key technologies proposed can reflect the actual network operation and can provide a reference for network optimization.The software system can also take full advantage of Spark parallel processing capabilities to analyze massive data.Performance analysis of the system can provides reference information for future data processing.