Joint Transmit and Receive Beamforming Design for Joint Multi-User Interference and Self-Interference Rejection for Full-Duplex Sub-Connected MU-mMIMO

Full-duplex (FD) radios have the potential to double the spectral efficiency of communication systems by transmitting and receiving on the same time and frequency slot. However, realizing this potential gain is challenging due to the strong selfinterference (SI) between the transmit and receive arrays. In massive MIMO (mMIMO), traditional cancellation techniques such as analog cancellation become unfeasible due to the required hardware complexity. The large number of antenna elements in mMIMO enables spatial processing techniques, making spatial SI suppression a promising approach for FD communications. This paper proposes Regularized Joint Linearly Constrained Minimum Variance Beamforming (RJLCMV), a novel alternating method for spatial SI cancellation for FD subconnected multi-user mMIMO based on regularized LCMV. The beamformer design problem is not jointly-convex between the transmit and receive array beamformers and is susceptible high levels of self-nulling with standard alternating optimization algorithms. The proposed algorithm uses diminishing regularization to mitigate the self-nulling effect while still providing total SI isolation. We study the cases in which total SI rejection can be achieved and demonstrate that with a measured SI channel, a duplexing gain of 1.93 can be achieved with RJLCMV in the fully constrained case in which both multi-user interference and SI are nulled, supporting as many users as there are per antenna element in each subarray.