Iterative Channel Estimation and Decoding for Coded MIMO System in Unknown Spatially Correlated Noise

We present an iterative channel estimation and decoding scheme for a coded system through multi-input multioutput (MIMO) Rayleigh block fading channels in spatially correlated noise. An expectation-maximization (EM) algorithm is utilized to find the maximum likelihood (ML) estimates of the channel and the spatial noise covariance matrix, based on which the soft information of coded symbols is computed and sent to the error-control decoder. Extrinsic information produced by the decoder is fed back to the channel estimators to refine the channel estimation. Several iterations are performed between the channel estimators and the decoder. We also develop a deterministic ML channel estimator that uses soft information from the decoder to update the channel and the noise covariance estimation. There is negligible difference between the performance with the EM-based channel estimation and that with the true channel at bit error rate of 10.