Research On Network State Quality Evaluation And Prediction Of LTE-M Vehicle Ground Wireless Communication System

Nowadays,urban rail transit has the characteristics of high speed and high density,which requires the wireless communication system to be safe,efficient and high transmission rate.In practical application,train control system often causes train delay due to poor network quality or external interference.Therefore,identifying the causes of network failure,realizing real-time evaluation of network status quality and predicting the future network quality can improve the performance of communication system to some extent and ensure the safe transmission of train control information.The network state quality evaluation and prediction of LTE-M vehicle ground wireless communication system are mainly studied in this thesis.Firstly,the research background and significance of this topic are introduced,and the research status of urban rail transit communication system at home and abroad is analyzed.The business requirements and transmission performance requirements for LTE-M in the industry are listed.According to LTE technology characteristics,the key performance indexes which can directly reflect the network state quality are selected and the factors affecting the network state quality are analyzed.Through the optimization analysis of station gauge,station spacing and other parameters,as well as the performance comparison between scheduling algorithm,switching algorithm and other algorithms,combined with the system requirements and the actual scene,the simulation scene suitable for LTE-M communication system is built based on ns-3network simulator.Then collect the KPI data for statistical processing,and analyze the network performance.Based on the completion of the scene,combined with the analysis of the network quality factors,then configure the relevant parameter settings.Simulate equipment aging,or power failure,antenna failure,interference and other scenarios caused by sudden situation.Based on the support vector machine to optimize the parameters,the optimal parameters are obtained and the classification accuracy is verified.Then the KPI data is classified to realize the evaluation of the network state quality and give the fault solving measures.Finally,the deep learning prediction method is used.After that,we make full use of the correlation between KPI data and time series to achieve the prediction and optimization of the future network state quality,so as to find the potential failure risk,realize the abnormal performance early warning and trend early warning,and improve the system performance.