A robust precoder design based on channel statistics for MIMO-OFDM systems with insufficient cyclic prefix

In this paper, we study the precoder design problem for MIMO-OFDM systems with insufficient cyclic prefix (CP) using statistical channel state information (CSI) at the transmitter. In our previously proposed channel independent precoding scheme, an interference nulling based precoding structure is studied, which can assist to either cancel interblock interference (IBI) or separate the subspace occupied by IBI from the subspace used for information symbols. In this work, we consider a robust precoder design which combines statistical CSI at the transmitter and the interference nulling based precoding structure to gain a better performance compared to the channel independent precoding scheme. Statistical channel state information at the transmitter in the form of covariance matrix of the MIMO-OFDM channel matrix is utilized to perform the robust precoding. The design criterion for the precoder design is to minimize the maximum average mean squared error (MSE) of all the transmitted symbols during one OFDM block interval. To achieve this goal, the precoder design exhibits a similar procedure to the optimized design framework for MIMO transceivers for either perfect or statistical CSI with no interference effect, but with a distinct equivalent channel. Since the average MSEs are set equal using this precoder, the BER performance is better than our previously proposed channel independent precoder and the simulation validates this result.