Fast spatial correlation acquisition for hybrid precoding using sequential compressive sensing

This paper addresses the critical problem of high overhead in channel information acquisition for hybrid precoding in massive multiple-input multiple-output (MIMO) system. With the employment of large-scale massive antennas in the next generation 5G wireless system, the high overhead in the estimation of channel state information (CSI) for the precoding has been regarded as one of the main challenges that prevent its implementations. On the other hand, other efficient hybrid precoding schemes such as beam sweeping based two stage precoding have limited performance compared with the complete CSI based precoding algorithm. This paper proposes a novel technique to quickly acquire the channel spatial correlation of the large number of antennas, which can be used for hybrid precoding design. This technique can greatly reduce the training overhead while maintain the capability of generating highly user-specific analog beamforming. The simulation results show that the proposed technique can achieve similar performance with the traditional complete CSI based precoding but only require even less reference signal (RS) symbols than the beam sweeping scheme.