Channel Estimation for Near-Field Line-of-Sight XL-MIMO Communications Using Geometric Prior

This letter investigates the channel estimation for near-field line-of-sight (LoS) extremely-large multiple-input-multiple-output (XL-MIMO) communications, where both the transceivers are equipped with uniform linear arrays (ULAs). For near-field LoS channels within millimeter wave and sub-teraherz frequency bands, electromagnetic waves typically propagate with little diffraction or scattering, implying the propagation direction of each plane-wave component remains almost invariant under the Fourier plane-wave series expansion. This invariance indicates a strong correlation between the propagation directions of transmitted and received plane-wave components, called geometric prior in this letter. Initially, we utilize the wavenumber-domain sparsifying basis to decompose the channel into several plane-wave components. Then, we deduce the potential locations of non-zero entries in the wavenumber-domain channel matrix for any relative orientation angle between the ULAs. Subsequently, a two-stage channel estimation framework is further proposed. Specifically, the first stage aims to derive the orientation angle, and the second stage performs compressed sensing within a reduced search space. Simulation results are provided to validate the robustness of our proposed angle and channel estimation.