Asymptotic Behavior Analysis and Performance Optimization in Full Duplex Massive MIMO

In this paper, we consider rate maximization of a single full-duplex (FD) massive MIMO base station (BS) with multiple downlink (DL) and uplink (UL) users. The BS applies transmit precoders on the DL transmission. These precoders are designed to reduce the multiuser interference among DL transmissions and to reduce the self-interference (SI) level from the DL transmissions at the BS UL receiving antennas. The self-interference is reduced by zero- nulling the DL transmissions at some of the UL receiving antennas. We derive lower bounds on the achievable DL and UL capacities. Based on the derived bounds, we optimize over the ratio of the transmit to receive antennas at the base station to maximize the achievable capacities. We also optimize over the portion of UL receive antennas that are zero-nulled by the DL transmit precoders to limit the effect of SI on the UL transmissions. Two different formulations of the rate region maximization problem are presented. The first formulation, which maximizes the UL rate for a given DL rate, is proved to be a convex optimization problem; for this formulation, an exact solution for the optimization problem is derived. The second formulation, which maximizes the DL rate for a given UL rate, is proved to be non-convex; therefore, the dual problem is formulated and solved. Numerical results validate the derived DL and UL bounds as well as the proposed rate maximization solutions.