Semi-blind MIMO Channel Estimation with Orthogonal Inputs

Semi-blind channel estimation based on the orthogonal procrustes (OP) algorithm is studied for multipleinput multiple-output (MIMO) systems with structured orthogonal inputs design. The proposed scheme is based on decomposition of channel matrix H as a whitening matrix R and a unitary rotation matrix Q. A blind estimation scheme for the matrix R based on eigenvalue decomposition (EVD) approach is also proposed along with an efficient OP algorithm for the matrix Q. This results in a substantially improved accuracy of the channel parameterization. Furthermore, by using such optimum rotation matrices, which are parameterized a fewer number of parameters, the proposed method can achieve low computational complexity. Computer simulations are included to validate that the proposed schemes achieve excellent performance, outperforming conventional ML method, and 1-3 dB improvement of mean squared error (MSE) would result if the same length of training sequence is used.