Research On 5G Field Strength Prediction Model For Marshalling Yard Scenarios

5th Generation Mobile Communication System(5G)technology research has promoted the development of the railway mobile communication system.The railway mobile communication system is transitioning smoothly from Global System for Mobile Communications-Railway(GSM-R)to the railway 5G,which includes the Fifth Generation of Mobile Communications for Railway(5G-R)and public network 5G applications in the railway industry.Railway 5G not only needs to undertake a variety of business requirements under a larger bandwidth,but also faces challenges brought by new scenarios,new frequencies and new networking modes.The marshalling yard railway 5G communication system requires the wireless network to provide stable coverage to complete a variety of special services.Path loss prediction is a key step in the construction of the railway 5G communication system.Therefore,considering the lack of a corresponding path loss prediction model in the marshalling yard scenario in the current 5G frequency band,the path loss prediction model in the marshalling yard scenario in this paper can help network planners in link budgeting,coverage prediction,system performance optimization and base station(BS)location selection work,so that the wireless network can meet the actual needs.Under this background,this paper researches path loss prediction model in the marshalling station scenario.It mainly contributes to the following aspects:Firstly,an empirical model of path loss is established based on a large amount of measured data in the marshalling yard scenario.Based on the characteristics of the base station directional antenna and pattern calibration,the position of the breakpoint in the antenna propagation area is proposed.Based on the breakpoint segmentation,a large-scale statistical model for 5G marshalling yards is established.Secondly,a path loss machine learning model is established based on a large amount of measured data and back propagation neural network(BPNN)in the marshalling yard scenario.The corresponding environmental features are represented by map rasterization,and the validity of the environmental features is verified,and then a large-scale machine learning model for 5G marshalling yard is established by combining the system features and environmental features to build a dataset.Eventually,based on the evaluation index of regression model,the generalization ability of path loss machine learning model is verified and evaluated,and complete the comparative evaluation of the existing model and the self-built model in the marshalling yard scenario,and give the wireless coverage prediction model of the scene of the marshalling station.In summary,this paper focuses on the path loss prediction model in the marshalling yard scenario,provides a path loss prediction model for network planning under the marshalling yard,and enriches the theoretical research in the marshalling yard scenario.