Hierarchical codebook design for massive MIMO

The massive Multiple-Input Multiple-Output (MIMO) is an emerging research field, since the more antennas the transmitters or receivers equipped with, the higher spectral efficiency and link reliability the system can provide. Due to the limited feedback channel, precoding and codebook design are important to exploit the performance of massive MIMO. In this paper, we propose a novel hierarchical codebook with the Fourier-based perturbation matrices as the subcodebook and the Kerdock codebook as the main codebook, which has reduced storage and search complexity due to finite alphabet. Moreover, to further reduce the search complexity and feedback overhead without noticeable performance degradation, we use an adaptive selection algorithm to decide when to use the subcodebook. Simulation results show that the proposed codebook has remarkable performance gain compared to the conventional Kerdock codebook, without significant increase in feedback overhead and search complexity.