Learning-Based Prediction Method for Radio Wave Propagation Using Images of Building Maps

In this letter, we propose a learning-based method that uses building map images as spatial data to predict radio wave propagation. We previously proposed a prediction method that uses images around a transmitter (Tx), a receiver (Rx), and their midpoint. The method was accurate but was missing regions between the Tx and the Rx when the distance between the two increased. In this letter, we modify the method to include the entire area between the Tx and the Rx, as well as their surrounding regions in a single image. We present a method of using the dataset generated from the measured data and evaluate the extent to which the proposed method improves the accuracy of predicting radio wave propagation.