Pilot Signal and Channel Estimator Co-Design for Hybrid-Field XL-MIMO

TL;DR

This work tackles hybrid-field channel estimation for XL-MIMO by jointly designing pilot signals and estimators to recover a sparse mixed angular/polar domain channel that includes LoS, far-field, and near-field components. The authors formulate a mutual coherence minimization problem and solve it with an ADMM-based pilot design, producing sensing matrices that enable reliable CS recovery. Building on the optimized pilots, they propose a two-stage estimator that first solves for the LoS channel and then employs Bayesian matching pursuit (BMP) to recover the hybrid-field scattering channel, with variants for both with and without prior channel knowledge. Simulation results show that the proposed co-design significantly outperforms conventional HF-CS methods in NMSE and spectral efficiency, closely approaching Genie-aided performance at moderate-to-high SNR, demonstrating practical impact for 6G XL-MIMO systems.

Abstract

This paper addresses the intricate task of hybrid-field channel estimation in extremely large-scale MIMO (XL-MIMO) systems, critical for the progression of 6G communications. Within these systems, comprising a line-of-sight (LoS) channel component alongside far-field and near-field scattering channel components, our objective is to tackle the channel estimation challenge. We encounter two central hurdles for ensuring dependable sparse channel recovery: the design of pilot signals and channel estimators tailored for hybrid-field communications. To overcome the first challenge, we propose a method to derive optimal pilot signals, aimed at minimizing the mutual coherence of the sensing matrix within the context of compressive sensing (CS) problems. These optimal signals are derived using the alternating direction method of multipliers (ADMM), ensuring robust performance in sparse channel recovery. Additionally, leveraging the acquired optimal pilot signal, we introduce a two-stage channel estimation approach that sequentially estimates the LoS channel component and the hybrid-field scattering channel components. Simulation results attest to the superiority of our co-designed approach for pilot signal and channel estimation over conventional CS-based methods, providing more reliable sparse channel recovery in practical scenarios.

Figures (8)

• Figure 1: An illustration of a downlink XL-MIMO communication system, where the channel components are classified into two types based on the Rayleigh distance: i) far-field and ii) near-field.
• Figure 2: An illustration of the distinction between the near-field region and the far-field region, which is determined by the Rayleigh distance.
• Figure 3: A schematic diagram of the near-field channel model
• Figure 4: NMSE versus SNR with an LoS path and $L=4$ scattered paths for four hybrid-field channel estimation methods: 1) HF OMP; 2) HF SD-OMP; 3) BMP w/ CSI; 4) BMP w/o CSI.
• Figure 5: NMSE versus pilot signal length with an LoS path and $L=4$ scattered paths at an SNR of 10 dB for four hybrid-field channel estimation methods: 1) HF OMP; 2) HF SD-OMP; 3) BMP w/ CSI; 4) BMP w/o CSI.