A random-vector-based beamforming scheme for millimeter-wave MIMO systems at low SNR

A random-vector-based beamforming scheme for Millimeter-Wave (mm-Wave) MIMO systems with asymptotic optimal performance at low signal-to-noise ratio (SNR) is proposed in this paper. Through transmitting random vectors from transmitter (TX) to receiver (RX) then back to TX, the proposed scheme can obtain the estimation of channel auto-correlation with low complexity operations and then obtain the nearly optimal transmit beamforming vector in baseband processor by multiplication and normalization quickly. Then the corresponding receive beamforming vector is acquired at RX by transmissions. The developed schemes including the Gaussian-random-vector-based scheme and the binary-random-vector-based scheme can reduce the impact of additive white Gaussian noise (AWGN) in theory to achieve nearly optimal equivalent channel gain. Through simulations, we can see that our scheme has advantages over iterative beamforming scheme especially at low SNR with better convergence which may be useful in secret communication occasion where high-power beamforming training is prohibited.