The Performance Analysis on Channel Estimation in Millimeter-Wave Communication and Their Challenges

mmWave MIMO systems offer enhanced better gains of antennas and the less difficulty level of hardware. The task of estimating channels is arduous owing to the huge amounts of transmitters/receivers at Base Station (BS), while the IRS integrates passive reflective elements without active transmitters/receivers. Initially, the computation of channel and optimal hybrid pre-coder/combiner design relay on Compressive Sensing (CS) algorithms, specifically the Orthogonal Matching Pursuit (OMP) scheme. OMP introduces high computational complexity, particularly when dealing with high-dimensional sparse signals to. To cope with this issue, researcher proposed Approximate Message Passing (AMP) and Sparse Bayesian Learning (SBL) techniques. While the SBL algorithm exhibits robust performance, it requires a lot of time in terms of complexity. The AMP and LAMP networks ensure performance with certain constraints on the measurement matrix, limiting their applicability in solving CS problems. For non-i.i.d. Gaussian matrices and i.i.d. Gaussian matrices, the AMP, LAMP network, SBL-EM, and SBL-GAMP do not exhibit convincing performance. Therefore, to get convincing estimation for non-i.i.d. Gaussian matrices and i.i.d. Gaussian matrices, we present a Vector AMP (VAMP) to SBL in addition to Expectation Maximization (EM) method. The VAMP-based SBL algorithm integrates the SBL structure with the robust characteristics of VAMP, demonstrating improved performance, particularly under various measurement matrices and a GM prior. The final results exhibit the outstanding performance of the proposed VAMP-SBL over the existing state-of-the-art approaches.