Skip to Main Content
In this paper, a novel signal-perturbation-free (SPF) approach is presented for frequency-selective multiple-input-multiple-output orthogonal frequency division multiplexing channel estimation. First, an efficient transmit scheme, which bears partial information of the correlation matrix of the transmitted signal called SPF data, is proposed for the cancellation of signal-perturbation error at the receiver. A detailed transmit structure is designed to implement the SPF scheme, which is then employed along with linear prediction (LP) to derive a new semiblind channel-estimation algorithm. It is shown that the new transmit scheme can completely cancel the signal-perturbation error in the noise-free case while being able to sufficiently suppress the perturbation error in noisy conditions. It is also shown that the SPF data needs only to be transmitted over a small number of subcarriers, and its overhead to the overall transmission is negligible as compared with regular pilot signals. Computer simulations show that the proposed SPF solution significantly outperforms the LP semiblind method without using the proposed transmit scheme as well as the least square method in terms of the mean-square error of the channel estimate.