Pilot-Efficient Channel Estimation for Near-Field XL-MIMO Under Spatially Non-Stationary Channels

Abstract

Extremely Large Multiple-Input Multiple-Output (XL-MIMO) systems are considered an important technology for future wireless communication because their large antenna apertures can provide high spatial resolution, beamforming gain, and spectral efficiency. However, the large number of antenna elements significantly increases the pilot overhead required for accurate channel state information (CSI) acquisition. In addition, spatial non-stationarity causes different portions of an XL-MIMO array to experience different channel characteristics, making conventional channel estimation techniques less effective. This paper proposes a Pilot-Efficient Channel Estimation framework for Near-Field XL-MIMO Under Spatially Non-Stationary Channels. The proposed approach exploits the spatial characteristics of the near-field channel and selectively acquires pilot information from informative antenna sub-regions instead of relying on uniform pilot transmission across the complete array. The received pilot observations are subsequently used to reconstruct the channel response across the XL-MIMO aperture. A simulation framework is developed to evaluate the proposed approach under different antenna array sizes, user distances, signal-to-noise ratios, and spatial non-stationarity conditions. Performance is analyzed using channel estimation mean square error (MSE), bit error rate (BER), pilot overhead, spectral efficiency, and computational complexity. The proposed framework aims to reduce pilot requirements while maintaining reliable CSI estimation and communication performance, providing an efficient approach for practical near-field XLMIMO systems.