Near-Field Channel Estimation and Environment Mapping: Localization of Reflectors and Scatterers

Environment sensing and mapping are crucial for integrated sensing and communication (ISAC), where the ultra-large antenna array (ULAA) improves resolution but also induces near-field effects. Near-field channel estimation is challenging, especially due to spatial non-stationarity (SNS) from directive reflection. In particular, reflection leads to (i) partial ULAA illumination, which causes SNS and model mismatch, and (ii) varying reflection points across antenna elements, which causes ambiguity in environment mapping. To address these issues, we incorporate geometry constraints (GCs) of propagation mechanisms into the space-alternating generalized expectation-maximization (SAGE) framework. The proposed GC-SAGE can estimate SNS channels and locate both the reflecting surfaces and scatterers, where ray-based simulations validate its effectiveness for accurate near-field environment mapping.