Adaptive Near-Field Channel Estimation for XL-MIMO Under Blockage

Abstract

The deployment of large-scale arrays over high-frequency bands extends the near-field region and enables beam focusing to enhance received power. However, high-frequency signals are sensitive to pervasive obstacles in near-field environments, inducing unknown non-uniform channel gain variations across the antenna aperture and posing a critical challenge for channel estimation (CE). To address this challenge, this letter investigates near-field CE under blockage. To enable obstruction-case CE from compressed measurements under hybrid combining, the complex gain of each path is represented as a spatially varying vector across the antenna aperture, capturing diffraction-induced non-uniform amplitude while maintaining compatibility with unobstructed cases. A two-stage estimation algorithm is then proposed that sequentially identifies the angle-distance parameters of each path via a polar-domain dictionary, and recovers the spatially varying complex gain profile via the discrete cosine transform (DCT) basis. Furthermore, an F-test-based criterion is presented to adaptively determine the path number. Simulation results demonstrate substantially lower estimation error than traditional near-field schemes under comparable complexity.