Joint Inter-and-intra-multiplexing and Hybrid Beamforming for Terahertz Widely-spaced Multi-subarray Systems

Terahertz (THz) communications with multi-GHz bandwidth are envisioned as a key technology for 6G wireless systems. While suffering from huge propagation loss, large arrays of sub-millimeter wavelength antennas can be realized in ultra-massive (UM) MIMO systems to overcome the distance limitation. However, channel sparsity and low spatial degree-of-freedom of the THz channel limit the spatial multiplexing gain and hence the achievable rate. In this paper, a widely-spaced multi-subarray (WSMS) hybrid beamforming architecture is proposed for THz UM-MIMO systems to improve the multiplexing gain. In each subarray, the antennas are critically-spaced to utilize the inter-path multiplexing and beamforming gains, while the subarrays are widely-spaced to harvest the novel intra-path multiplexing gain through exploiting the phase differences over the subarrays. By exploring the THz channel peculiarity, a dominant-LoS-relaxation (DLR) method is proposed to balance the multiplexing and beamforming, and a block-diagonal vectorization-based (BD-VEC) algorithm is developed to solve the hybrid beamforming problem. Extensive simulation results demonstrate that the multiplexing gain is improved by a factor of the number of subarrays in the THz WSMS system. The low-complexity DLR method maximizes the achievable rate near-optimally. Compared to existing hybrid beamforming algorithms, the BD-VEC algorithm achieves higher spectral efficiency with substantially lower computational complexity.