Deep Learning-Based 5G SINR Prediction using Urban DEM data

Accurate prediction of 5G SINR (Signal-to-Interference-and-Noise Ratio) in urban environments is essential for the effective operation of 5G networks. To date, SINR prediction has predominantly been performed through channel modeling using statistical approaches. However, these methods have limitations in adequately accounting for geospatial information, such as buildings, in urban areas. In this study, we propose a deep learning model that predicts the SINR of user equipment (UE) at specific points by fully incorporating geospatial information from digital elevation model (DEM) data and 5G base station (BS) locations. The prediction results demonstrate that our model provides more accurate predictions compared to existing methods and closely aligns with actual measurements.