Blind Channel Subspace Estimation for Massive MIMO with Hybrid Beamforming

Massive MIMO is one of the key technologies for 5G wireless, for which hybrid beamforming architecture can be used to reduce hardware cost and power consumption, while still achieving huge spectral efficiency. But the hybrid beamforming architecture complicates the channel estimation because the number of radio frequency (RF) chains is far less than the number of receiving antennas. In this paper, we propose a blind scheme to estimate the subspace spanned by the column vectors of the MIMO channel assuming no training sequences. The key is to apply different analog beamformers to obtain a set of covariance matrices of the outputs of the analog-to-digital converters (ADC). Using the matrices, we propose two algorithms to estimate the column space of the channel matrix. Simulation results verify the effectiveness of the proposed algorithms.1