Improvement of Deep Learning-Based Reference Signal Received Power Prediction for LTE Communication System

Abstract

Recently, in our previous work [3], we have proposed the deep learning-based reference signal received power (RSRP) prediction for LTE communication system. However, in this work, the output of DNN is path loss error instead of RSRP. Moreover, our deep neural network is improved by increasing the number of features, such as 3-D antenna gain, digital elevation model (DEM). The path loss model is also developed by using the clustering technique. The results show that the dominant prediction testing, non-dominant prediction testing can provide the RMSE around 3.7 dB, and 4.9 dB, respectively.