Hybrid Beamforming Based on Dictionary Learning for Millimeter Wave MIMO System

Hybrid beamforming in transmitter and receiver can improve the spectral efficiency of millimeterwave (mmWave) multiple-input multiple-output (MIMO) communication system significantly, and can reduce the complexity of communication system. However, when the number of transmitted data streams is large, the existing designing schemes of hybrid beamforming, such as sparse approximation by orthogonal matching pursuit(OMP) algorithm, suffer from performance degradation. In this paper, we propose dictionary learning (DL) algorithm to perform the design of hybrid beamforming for mmWave MIMO system. Simulation result shows that the spectral efficiency of hybrid beamforming based on DL approaches that of optimal beamforming without constraint, which is far superior to the spectral efficiency of hybrid beamforming based on OMP. Moreover, the convergence property of the proposed algorithm is also verified.