Two-Stage Beamspace MUSIC-Based Near-Field Channel Estimation for Hybrid XL-MIMO

Abstract

In extremely large-scale multiple-input multiple-output (XL-MIMO) systems, channel estimation poses a key challenge due to the introduction of the unknown distance parameter in near-field scenarios. We propose a beamforming codebook that includes pre-compensated distances, which allows the application of the traditional beamspace multiple signal classification (BMUSIC) to near-field channel estimation. To determine the optimal pre-compensation distance, we introduce three strategies: Maximizing the correlation integral (MCI), maximizing the minimum correlation (MMC), and exceeding the minimum correlation threshold (EMCT). In addition, we develop a two-stage BMUSIC algorithm and a switch transformation design to further reduce the time-intensive 2-dimensional (2D) search processes and avoid the overlaps of multiple coherent paths. Simulation results confirm that the proposed method not only diminishes computational complexity but also notably outperforms existing methods in terms of estimation accuracy.