POPT: Physics-Informed Deep Learning for Phase-Only Positioning in Distributed MIMONew

This paper develops a physics-informed deep learning framework for single-snapshot 3D phase-only positioning in narrowband, phase-coherent distributed MIMO networks. Its electromagnetic propagation model considers a line-of-sight path and a specular ground reflection under a two-ray channel. The method addresses the inability of narrowband single-snapshot observations to resolve the two propagation components in delay.