Skip to Main Content
In this paper, a new blind channel estimator for multiple-input multiple-output orthogonal frequency-division multiplexing (MIMO-OFDM) systems is proposed. Based on the properties of the underlying orthogonal space-time block code (OSTBC), a novel subspace model is devised. The proposed technique consists in a closed-form approach in which the channels are estimated from the principal eigenvector of an extended sample covariance matrix of the received data. It is capable to take advantage of the finite delay spread assumption of the channel to estimate a lower number of channel parameters in the time rather than in the frequency domain. The coherent processing across all the subcarriers not only yields a considerable noise reduction and improved estimation accuracy, but also guarantees uniqueness of the channel estimates under certain conditions and offers a substantially reduced computational complexity.