A Full-Dimension Channel Estimator for mmWave/THz MIMO Using Weighted Atomic Norm Minimization

Due to the high path loss of millimeterwave/terahertz (mmWave/THz) frequency wave propagation, it necessitates fine-grained beamforming with super-large antenna arrays to compensate for transmission gains, which is naturally attributed to accurate channel state information. In this paper, a super-resolution channel estimation method is proposed for the full-dimension (FD) mmWave/THz communication system, where both the base station (BS) and the receiver pack uniform planar arrays (UPAs). Benefiting from the strong line-of-sight property of the mmWave/THz channels, the angular block prior is taken into account by the proposed channel estimators to improve the estimation accuracy and/or reduce the training overhead. The off-grid FD channel estimation problem is formulated based on weighted atomic norm minimization, which enhances the inherent sparsity in the continuous elevation-and-azimuth angular block domains. Simulation results show that the proposed methods can achieve considerable performance gains, as compared with existing both on-grid and off-grid approaches.