Sequential Convex Programming for Full-Duplex Single-User MIMO systems

This paper proposes two Sequential Convex Programming (SCP) algorithms, namely Difference of Convex functions (DC)-based and Sequential Convex Approximations for Matrix-variable Programming (SCAMP), for solving the non-convex matrix-variable sum-rate maximization problem in Full-Duplex (FD) Single-User Multiple-Input-Multiple-Output (SU-MIMO) systems. The two proposed algorithms result in different approximations of the objective function and hence, depending on the environment, one may be favorable than the other. Numerical results show that SCP can significantly increase the sum-rate over existing techniques for the SU-MIMO scenario. In particular, for the SU-MIMO scenario, the DC-based algorithm outperforms the SCAMP.