Bayesian Channel Estimation in Multi-User Massive MIMO With Extremely Large Antenna Array

• Published in: IEEE transactions on signal processing 2021, Vol.69, p.5463-5478
• Main Authors: Zhu, Yifan, Guo, Huayan, Lau, Vincent K. N.
• Format: Article
• Language: English
• Subjects: Algorithms; Antenna arrays; Antennas; Bayes methods; Bayesian analysis; Channel estimation; Extremely large antenna array; Hidden Markov models; Markov chains; Massive MIMO; Message passing; MIMO communication; MISO (control systems); Scattering; Sparsity; Statistical inference; structured sparsity; Visibility; Wave fronts
• ISSN: 1053-587X; 1941-0476
• DOI: 10.1109/TSP.2021.3114999

We investigate wideband uplink channel estimation for a multi-user (MU) multiple-input single-output (MISO) OFDM system, in which the base station (BS) is equipped with an extremely large antenna array (ELAA). The existing compressive sensing massive multiple-input multiple-output (MIMO) channel estimation approach with a traditional sparsity promoting prior model becomes invalid in the ELAA scenario due to the spatial non-stationary effects caused by the spherical wavefront and visibility region (VR) issue. We therefore propose a new structured prior with the Hidden Markov Model (HMM) to promote the structured sparsity of the spatial non-stationary ELAA channel. Based on this, a Bayesian inference problem on the posterior of the ELAA channel coefficients is formulated. In addition, we propose the turbo orthogonal approximate message passing (Turbo-OAMP) algorithm to achieve a low-complexity channel estimation. Comprehensive simulations verify that the proposed algorithm has supreme performance under spatial non-stationary ELAA channels compared to various state-of-the-art baselines.