Taming the complexity of mm-wave massive MIMO systems: Efficient channel estimation and beamforming

Abstract

Massive multiple input multiple output (MIMO) millimeter wave (MMW) communications allow for a compact implementation and efficient beamforming. In this paper we focus on the problem of estimating the massive MIMO channel in a code modulated path sharing multi-antenna (CPMA) architecture, when a limited number of radio frequency (RF) chains is available, and beamforming is performed through a combination of analog and digital signal processing. Antennas are organized in uniform planar arrays (UPAs). The proposed solution is based on estimation of a channel sub-matrix and exploits properties of the UPA model. In particular, we first organize UPA indices in a way that simplifies further processing, then we propose a training sequence that minimizes the number of RF chains. Lastly, rather than estimating the subchannel matrix itself we estimate parameters of its entries by an efficient post-processing technique based on a four dimensional Fourier transform that exploits properties of the UPA model.