Analog Beamforming and Combining Based on Codebook in Millimeter Wave Massive MIMO Communications

Analog beamforming at transmitter and analog combining at receiver can be designed based on a hierarchical codebook, which consists of a small number of low-resolution codewords covering wide angle at top level and a large number of high-resolution codewords offering high directional beamforming gain at bottom level. Although high-resolution codewords are preferred, it takes more time for channel training. In this paper, weighted sum-rate maximization for millimeter wave (mmWave) massive MIMO communications is investigated by jointly considering the duration for channel training and the receiving signal-to-noise ratio (SNR). An algorithm is proposed to design the analog beamforming and analog combining based on the hierarchical codebook. At each iteration, the channel gain of the dominant path is estimated. The level of the hierarchical codebook achieving the weighted sum-rate maximization is predicted and then fed back to the transmitter. If the predicted level is reached, the transmitter and the receiver stop channel training and begin data transmission using the designed analog beamforming and analog combining. Simulation results show that the proposed algorithm outperforms multi-sectional search, where the former can achieve the weighted sum-rate almost twice of the latter at certain length of transmission block when the channel SNR is 20dB.