Channel estimation in massive MIMO-based wireless network using spatially correlated channel-based three-dimensional array

To more benefit from the massive multiple-input multiple-output (M-MIMO) technology for improving the channel estimation (CE) process, each base station (BS) must have accurate channel state information as effectively as possible. In this work, we address the CE process for the M-MIMO network, considering a more practical scenario where the channels are spatially correlated, as the spatial correlation (SC) strongly affects the performance of M-MIMO systems. Additionally, the SC relies on several factors, such as the BS’s array arrangement. Thereby, we investigate the SC effect over CE using different array arrangements, wherein the SC is described using a gaussian local multi-scattering (GLMS) model. In addition, the performance of the Bayesian Minimum Mean Square Error estimator is investigated for correlated and uncorrelated channels. Besides, using the GLMS model for the uniform linear array (ULA) arrangement and based on the Kronecker product (KP), we propose the GLMS model for the uniform planar array arrangement. We also address the channel hardening and favorable propagation for both arrangements. Furthermore, we propose the GLMS model for the uniform circular array (UCA) arrangement, where we drive a theoretical demonstration of the GLMS model for the proposed UCA arrangement. Moreover, we propose a lower complexity GLMS model for uniform cylindrical array arrangement by relying on the KP of the proposed model for UCA arrangement and the GLMS model for vertical ULA (V-ULA) arrangement. The system performance is evaluated using the normalized mean square error metric, where the simulation results are in view to affirm our mathematical analysis.