Radio Propagation Prediction Using Deep Neural Network and Building Occupancy Estimation

抄録

In this paper, we propose a radio propagation prediction method using machine learning and building occupancy estimation. There have been learning-based researches using aerial photographs and building occupancy images as spatial information. However, the availability of building occupancy images are often limited to urban areas. We assume the situation that only the aerial photographs are available, and aim to improve radio propagation prediction accuracy by estimating building occupancy images from the given aerial photographs.