Joint Channel, CFO, and Data Estimation via Bayesian Inference for Multi-User MIMO-OFDM Systems

In this paper, we propose a novel low-complexity Bayesian receiver design to jointly perform channel, CFO, and data estimation from observations subject to different CFO among users in MU-MIMO-OFDM systems. ICI due to CFO significantly reduces channel estimation accuracy under frequency-selective fading environments, making reliable communications difficult. To tackle this difficulty, a JCCE algorithm is designed based on BP. Our method uses a BG distribution as the prior distribution of the channel coefficient to capture its delay-domain sparsity, and a GM distribution as the prior distribution of the phase shift due to CFO to perform parallel search for the allowable range of CFO defined in the 3GPP standard by the number of mixture components. The proposed algorithm can further improve the accuracy of channel, CFO, and data estimation by treating the tentatively detected data symbols as extra pilots. The efficacy of the proposed method is confirmed by numerical studies, which show that the proposed method not only significantly outperforms the SotA methods with much lower computational cost but also approaches the performance of an idealized Genie-aided scheme.