Line-of-Sight Based Statistical 3D Beamforming for Downlink Massive MIMO Systems

In this paper, we investigate the sum rate performance of a downlink transmission scheme for massive multipleinput multiple-output (MIMO) systems with a two-dimensional (2D) antenna array at the base station (BS). Assuming Rician fading conditions, we derive some properties of the channel's line-of-sight (LOS) component. Using these properties and under the assumption of only LOS information at the BS, the optimal beamforming vector for each user as well as the main guidelines for user scheduling are derived to maximize an approximation of the ergodic sum rate. Based on the optimal beamforming vector and the guidelines, a statistical three-dimensional (3D) beamforming downlink transmission scheme exploiting only the LOS information of each user is proposed. Numerical results show that the proposed scheme can achieve considerable sum rates while requiring much less channel state information (CSI) overhead at the BS, and is more attractive than the traditional zero-forcing and match-filtering schemes when the CSI at the BS is imperfect.