Sparse Bayesian learning-based data-aided channel estimation in STTC MIMO systems

In this paper, we consider a sparse multipath representation of the multiple-input multiple-output (MIMO) channel matrix in terms of an overcomplete dictionary consisting of the basis spatial signature matrices which correspond to various directional cosines at the transmit and receive antenna arrays. Based on the sparse Bayesian learning (SBL) framework, we exploit the spatially sparse representation of the MIMO channel and develop a novel pilot-based channel estimation scheme for space-time trellis coded (STTC) MIMO systems. Further, we propose an enhanced SBL-based data-aided channel estimation technique utilizing the expectation-maximization (EM) framework. We demonstrate that this can be derived as an optimal minimum mean squared error (MMSE) channel estimate in the E-step followed by a modified path metric-based maximum likelihood (ML) STTC decoder in the M-step. We also derive the Bayesian Cramer-Rao bounds (BCRBs) for the SBL-based pilot and data-aided channel estimation schemes. Finally, we present simulation results to demonstrate the performance of the proposed techniques and validate the analytical bounds.