Beamspace Channel Estimation for 3D Massive MIMO Cellular System

In this paper, we consider 3D multi-user massive multiple-input multiple-output (MIMO) cellular system with a base station (BS) and a number of uplink (UL) and downlink (DL) single-antenna users. The BS is equipped with large-scale uniform planar antenna array (UPA). Exploit the beam characteristic of 3D massive MIMO channels, we obtain the low-dimension beamspace of each user based on the direction of arrival (DOA) and direction of departure (DOD) of the signal. Then, we transfer the signal processing procedure to the beam domain. According to the particular beamspace of each user, we partition users whose beamspaces are the same into the same group while the beamspaces between groups are disjoint. With the orthogonal beamspaces, the inter-group interference can be eliminated as the BS antennas tend to infinity, and users within the same group are separated by orthogonal pilots. Finally, the simulation part shows that either the estimation performance or the overhead for estimation, the proposed beamspace channel minimum mean square error (MMSE) estimation scheme outperforms the conventional full-dimension MMSE channel estimation.