Learning Blockage and Reflection Geometry for MIMO Beam Map Construction

Understanding the wireless environment is becoming increasingly critical as communication networks advance more complex like 6G and beyond. In the absence of environment maps, extracting geometric features from signal measurements serves as an effective alternative to facilitate network intelligence and performance. This paper proposes to employ the received signal strength (RSS) data, without any knowledge of the city map, to jointly construct the radio beam map and environmental geometry for a multiple-input multiple-output (MIMO) system. Unlike existing methods that only capture blockage features, we propose a virtual reflector model that captures the geometric features of both blockage and reflection. Reflective zones are formulated to identify propagation paths that follow the law of reflection, and the aggregated effect of these paths is learned. By modeling the geometric relationship, a physics-inspired deep learning (DL) framework that incorporates the blockage component, the reflection component, and the beamforming gain pattern is proposed, with interpretability in network design. Numerical experiments demonstrate that, in addition to reconstructing the environmental geometry, the proposed method can construct a more accurate beam map with 25%-47% accuracy improvement, and reduce 85% probing pilots for beam sweeping.