Long-term adaptive precoding for decision feedback equalization

We consider the communication of digital signals over a multiple- input multiple-output wireless channel, using a linear precoder at the transmitter and a non-linear decision feedback equalizer at the receiver. This receiver structure can exploit the signal constellation properties by using successive quantization and interference cancellation. Recently, optimal precoder designs have been found for a wide range of performance measures assuming that perfect channel- state information (CSI) is available. Herein, we propose a design taking CSI uncertainty into account by utilizing the first and second order statistics of the channel. The resulting precoder exhibits improved performance compared to similar methods based on long- term statistics.