Preamble design for channel estimation in MIMO-OFDM systems

A maximum-likelihood (ML) channel estimation and preamble design rules for the multiple-input-multiple-output (MIMO) channels are described when orthogonal frequency division multiplexing (OFDM) is employed with null subcarriers at both DC and high frequencies. The ML channel estimator in the time domain is derived assuming the knowledge of the maximum length of channel. To reduce the mean square error (MSE) of the estimation, three design rules are proposed: orthogonality between the preambles of different antennas, orthogonality between the circular shifted preamble sequences, and the condition that the number of subcarriers; is larger than the maximum length of channel multiplied by the number of transmit antennas. In addition, we prove that the use of Golay Complementary Sequence (GCS) for preamble limits the peak to average power ratio (PAPR) by 3dB although there are null subcarriers at high frequencies. Numerical results shows that the MSE of the proposed method approaches that of optimum ML estimation when the number of null subcarriers and maximum length of channel are small.