Bilnd identification and blind equalization of MIMO channels based on QR factorization

Traditional blind identification methods for SIMO channels, such as subspace method (SS) etc., are based on the eigenvalue decomposition (EVD) or singular value decomposition (SVD) of the received signals' correlation matrix, which needs more sample data. Recently, a QR factorization based blind identification algorithm for SIMO FIR channels was proposed, which has good performance and need not calculate the correlation matrix. In this paper, we extend the QR based algorithm to blind identify MIMO FIR channels, and prove the algorithm is valid when the sources are spatially correlated. Simulations show that QR based algorithms preserve advantages of fast convergence, low computational costs, and robustness to noise.