Semi-blind channel estimation for a space-time system using a combined MEA-PI and auto-tracking algorithm

A novel semi-blind estimation scheme is proposed for an uncoded space-time system that uses a multi-element array pilot integration (MEA-PI) algorithm with Hadamard orthogonal structure combined with an auto-tracking algorithm. Central to the new method is a real-time channel estimator that introduces a very low pilot symbol overhead during the training process. The method is also capable of tracking the time variation of the channel using both the training vectors and a decision feedback algorithm during data acquisition process. As a result, a robust method of estimating the channel coefficients is possible, enabling the method to be used in real-time for a time-varying channel with a relatively short coherence time. The bit-error probability performance of this semi-blind method is investigated for different system configurations using non-linear MMSE decoding. It is found that the new method shows very little degradation in performance compared with the case where perfect channel estimates are assumed.