Coverage Area Estimation Using a Multi-Branch 1D Convolutional Neural Network

In cellular networks, coverage estimation is critical for network planning and optimization. Traditional ray tracing models the propagation of radio waves but faces limitations in large-scale applications due to high computational cost. Deep learning-based methods also accurately predict signal propagation, but are limited in large-scale applications due to the data requirements. In this study, large-scale synthetic datasets are created with Uniform Theory of Diffraction (UTD) based ray tracing simulations to overcome this problem. The proposed method generates 2D coverage maps by analyzing direct, reflected and diffracted electromagnetic propagation paths in 3D digital terrain maps. The developed “Multi-Branch Coverage Estimation Network” aims to estimate the signal propagation accurately and efficiently. Experimental results show that the proposed model provides high accuracy in coverage estimation and works more efficiently than conventional methods. Thus, high quality coverage estimation without the need for real measurements is an important step in wireless network planning.