MSE based transceiver designs for bi-directional full-duplex MIMO systems

We consider a multiple antenna full-duplex (FD) bi-directional (point-to-point) communication system with a limited analog domain self-interference cancellation capability. The effect of the residual self-interference resulting from independent and identically distributed (i.i.d.) channel estimation errors and limited dynamic ranges of the transmitters and receivers is studied in the digital domain. We design transceiver matrices based on the minimization of sum mean-squared error (MSE) and the maximum per-node MSE optimization problems subject to individual power constraints at each node through an iterative alternating algorithm, which is proven to converge to at least a local optimal solution.