Millimeter Wave Beam Alignment with Symbol Based Beam Switching for Wideband Time-varying Channels

We consider compressive sensing (CS)-based beam alignment for time-varying millimeter wave (mmWave) MIMO channels. We propose a symbol-based beam switching (SBBS) approach with channel-probing beams being switched every OFDM symbol. We use the SBBS to measure second order statistics of the channel, which are used for sparse recovery. We formulate the problem for a wideband channel model with pulse shaping. We consider low signal-to-noise ratios (SNRs) typical of mmWave channels and different fading rates in our simulations. Simulations show that our proposed SBBS approach outperforms existing approaches designed for longer beam-switching times.

I. Introduction

MmWave communications employs a large number of antennas at the base station (BS) and user equipment (UE) to enable beamforming (BF) that can combat the high pathloss in mmWave bands. The resulting beams are narrow and there is a significant challenge to find the best directions in which to point the beams. This task is called beam alignment (BA). The mmWave channel typically exhibits a small number of scatterers [1], and hence has a sparse channel matrix in the angle domain. Compressive sensing (CS) based BA methods have been developed for this problem [2]–[12].