Channel Estimation and Hybrid Beamforming for Reconfigurable Intelligent Surfaces Assisted THz Communications

Terahertz (THz) communications are promising to be the next frontier for wireless network but suffer from severe attenuation and poor diffraction. To address these challenges, this paper integrates two state-of-the-art technology, i.e., massive multiple input multiple output (MIMO) and reconfigurable intelligent surfaces (RISs), into THz communication to stablish effective and stable connection. Owing to the passivity of the RISs, channel estimation remains an open problem, besides, traditional fully-digital beamforming (FDB) can not be applied in massive MIMO due to its high hardware cost. In sight of this, this paper develops a cooperative channel estimation procedure for RIS-assisted MIMO system via beam training that exploits the advantages of massive MIMO and poor scattering nature of THz channel. The whole procedure is developed based on cost-efficient hybrid beamforming (HB) architectures. Moreover, this paper designs two different codebooks for efficiently realizing the proposed estimation procedure via 3-tree hierarchical search. The two codebooks are referred to as tree dictionary (TD) codebook and phase shifter deactivation (PSD) codebook, with different advantages catering for various scenarios. Using the estimation information, this paper develops two closed-form transmission designs, i.e., direct allocation (DA) design and zero-forcing allocation (ZFA) design, for maximizing the overall spectral efficiency.