Gradient Descent Based Polarization Channel Estimation in Extremely Largescale MIMO Systems

Accurate channel model and channel estimation are crucial for achieving extremely large-scale multiple-inputmultiple-output (XL-MIMO) in 6G networks with ultra-high spectral efficiency. As the aperture of the antenna array increases dramatically in XL-MIMO scenarios, the electromagnetic propagation field will change from far-field to near-field. In this paper, we propose a gradient descent-based near-field XL-MIMO channel model (GD-SOMP), an algorithm that makes the discrete adoption points more continuous to improve the estimation accuracy to match the actual near-field XL-MIMO scenarios. Numerical simulation results show that the GD-SOMP algorithm outperforms the conventional near-field channel estimation algorithm in terms of mean square error performance.