Adaptive Simultaneous Orthogonal Matching Pursuit for mmWave Hybrid Beam Tracking

Hybrid analog/digital precoding is an important technique for next generation millimeter wave (mmWave) multiple-input multiple-output (MIMO) systems. The hybrid precoding can reduce RF chain number and power consumption without sacrificing spectral efficiency and a large number of mmWave antennas can be easily packed in small dimensions. Therefore, hybrid precoding can be easily applied to massive MIMO systems. However, the computational complexity of hybrid precoding algorithm is quite high. Thus, this paper aims to reduce the complexity by eliminating redundant computations in practical communication environments. In continuous drifting channel, the channel state information between two successive time indices are strongly related. Accordingly, the proposed algorithm predicts the redundant computations for generating RF precoder and baseband precoder matrices. The proposed algorithm also considers spatial movements of scattering paths by selecting neighboring beamforming basis vectors in codebook search. Simulation results show that the proposed predictive update condition and neighboring search techniques reduce up to 61.3% and 57.1% complexity for the precoder and combiner, respectively.