Structured Channel Estimation and Localization for Wideband Near-Field MIMO Systems

Abstract

In wideband near-field multi-input multi-output (MIMO) systems, spherical-wave propagation induces a chirp-like channel structure, yet the intrinsic line-of-sight (LoS) element-wise transmit–receive coupling structure is usually neglected for improving channel estimation and localization accuracy, and the non-line-of-sight (NLoS) parameter estimation has high computational complexity due to joint angle–range searches. Furthermore, significant cluster-level spatial dispersion under spherical-wave propagation renders conventional point-scatterer channel models inadequate for accurate channel characterization. To address these limitations, we propose a unified LoS–NLoS near-field channel model and a low-complexity two-stage fractional Fourier transform (FrFT)-based scheme for channel estimation and localization. In the first stage, we exploit the chirp-focusing property of the FrFT to enable low-complexity channel parameter estimation for both LoS and NLoS components. In the second stage, wideband diversity and element-wise transmit–receive coupling structure facilitate LoS localization, and LoS-assisted geometric constraints resolve the NLoS pairing ambiguity caused by separate transmit and receive processing. We further derive the Cramér–Rao lower bound (CRLB) under the proposed LoS–NLoS near-field channel model to reveal the fundamental performance limits. Simulation results show that the proposed scheme is close to the CRLB and remains robust and accurate under unified LoS–NLoS near-field channel model.