Abstract In this article, a novel semi-blind channel estimation scheme for Multiple-Input Multiple-Output (MIMO) communication system is studied and implemented for quasi-static Rayleigh fading channel, in which channel matrix H remains relatively constant throughout the block. Channel matrix H can be decomposed as a rotation matrix Q and down triangular matrix R. The triangular matrix R is estimated blindly using the QR-decomposition based Generalized Cholesky Decomposition method (GCD) of output covariance matrix, which exploits Independent Component Analysis (ICA) based blind source separation stochastic method, and Q is estimated from the orthogonal pilot symbols using QR-based novel approach to minimize the cost function. In this novel approach, orthogonal pilots can be decomposed as a deterministic Hermitian matrix and the upper triangular matrix using the QR-decomposition, and finally matrix Q can be estimated by using the back substitution method, which is presented in this paper. Simulations are demonstrated using the Alamouti space-time coded 2 transmitter antennas and different combinations of 2 and 6 receiver antennas to showcase the performance of the novel scheme as compared to the conventional Least Squares (LS) and Maximum a posterior (MAP) estimation schemes using the BPSK data modulation scheme. The results indicate that the novel scheme outperforms the other schemes and shows a considerably better result in terms of bit error rate (BER). Hence, the novel scheme is quite useful for solving a complex problem of semi-blind MIMO channel estimation by using the QR matrix decomposition technique. Further error analysis is presented in terms of the error covariance matrix by considering the noise for non-zero error case (practical case) as compared to zero-error case.
展开▼
