Semi-blind iterative joint channel estimation and K-Best Sphere Decoding for MIMO

摘要

An efficient and high-performance semi-blind scheme is proposed for Multiple-Input Multiple-Output (MIMO) systems by iteratively combining channel estimation with K-Best Sphere Decoding (SD). To avoid the exponentially increasing complexity of Maximum Likelihood Detection (MLD) while achieving a near optimal MLD performance, K-best SD is considered to accomplish data detection. Semi-blind iterative estimation is adopted for identifying the MIMO channel matrix. Specifically, a training-based least squares channel estimate is initially provided to the K-best SD data detector, and the channel estimator and the data detector then iteratively exchange information to perform the decision-directed channel update and consequently to enhance the detection performance. The proposed scheme is capable of approaching the ideal detection performance obtained with the perfect MIMO channel state information.