Low-Complexity Hybrid-Field Channel Estimation for XL-MIMO Systems With Asymmetrical Uplink and Downlink Transceivers

Asymmetrical transceiver architectures in extremely large-scale multiple-input multiple-output systems employ fewer receive than transmit radio frequency chains to reduce hardware costs. However, the resulting small-dimensional uplink observations pose significant challenges for hybrid-field channel estimation. To address these challenges with low computational complexity, this letter proposes a discrete fractional Fourier transform (DFrFT)-based two-stage Newtonized orthogonal matching pursuit algorithm. By exploiting channel sparsity in the DFrFT domain, the proposed algorithm formulates channel estimation as a peak detection problem in DFrFT spectra. Through spectral peak analysis, the relationship between peak positions and channel parameters is established, from which the DFrFT rotation angle range where spectral peaks concentrate is derived, thereby significantly reducing the search space. Within this range, a two-stage detection strategy is developed to achieve low computational complexity, while Newton-based refinement ensures gridless accuracy. Without requiring prior knowledge of path numbers, the proposed algorithm efficiently reconstructs complete downlink channels from partial uplink observations.