Efficient and Accurate Semiblind Estimation of MIMO-OFDM Doubly-Selective Channels

This paper proposes an efficient and accurate semiblind channel estimation technique for Multiple-Input Multiple-Output Orthogonal Frequency-Division Multiplexing (MIMO-OFDM) turbo receivers. Once the channel is estimated using a few pilots, a low order Kalman filter is employed to progressively predict the channel gains for the upcoming blocks. A Basis Expansion Model (BEM) channel estimation scheme is used to allow the channel to vary within a block to make the method compatible with fast fading radio channels. As the detected data symbols are iteratively used by the Kalman filter to enhance the estimation accuracy, the proposed method compares with iterative pilot-aided systems in terms of computational cost and competes in spectral efficiency with semiblind and blind estimation techniques in fast fading environments. The BER performance of the proposed estimation approach is 0.3 dB off the perfect CSI case, whereas the computational complexity is on the order of that of near-optimal pilot-assisted methods.