Image-Driven Spatial Interpolation With Deep Learning for Radio Map Construction

Radio maps are a promising technology that can boost the capability of wireless networks by enhancing spectrum efficiency. Since spatial interpolation is critical challenge to construct precise radio map, latest works have proposed deep learning (DL)-based methods. However, DL model achieves enough estimation accuracy for practical uses has not yet been established due complexity propagation characteristics. Therefore, we propose novel framework transforms problem into shadowing adjustment suitable DL-based approaches. We evaluate performance using real measurement data in urban and suburban areas show outperforms state-of-the-art models.