Parameter Identification of Signal Attenuation for High Speed Rail Communication System Using Grey Box Neural Network

A reliable connected digital railways must face a huge challenge of network and frequency planning in order to deploy a dedicated infrastructure for High Speed Rail (HSR). This paper investigates the parameter identification of signal attenuation for HSR in a free space propagation case. A neural network based optimization of Grey Box Neural Network (GBNN) is utilized to validate the parameter identification results. Satisfying results of minimum convergences of RMSE = 1.6371 and MAPE = 0.015 have been achieved. Furthermore, the ability of identified parameter in adjusting the dynamic scenario of the real signal attenuation measurement dataset is displayed when the distance is 118.45 meter and the path loss value is 116.82 dB. Through this applied neural network based technique, network designers are able to plan and design the best possible of network performances and frequency planning for HSR.