Two-stages Multi-subarray Hybrid Precoding for the Near-Field Broadband THz MIMO Systems

Terahertz (THz) communications with multi-gigahertz bandwidth, are generally accepted as a keystone innovation for sixth-generation (6G) wireless communications, nevertheless, their severe path loss demands highly directional beam gain from extremely large-scale antenna arrays (ELAAs). However, most of the existing hybrid precoding algorithms neglect the beam split effect in the spherical-wave propagation model, which results in gain degradation at large bandwidths. The main challenges of ELAA wideband precoding include the recovery of near-field channel information, the exponential increase in computational complexity owing to the large number of antennas, and the effect of beam splitting caused by the highly directional beams and wide accessed bands. In this study, a two-stage multi-subarray wideband hybrid precoding algorithm is proposed. This algorithm improves the antenna gain and mitigates the beam split effect of the spherical-wave propagation model by exploiting a non-uniform subarray architecture (NUSA). For the proposed algorithm, the wideband hybrid precoding problem was decomposed into two subproblems. The first subproblem is near-field channel simplification. Remarks are proposed to optimize the antenna gain of the NUSA for a uniform linear array (ULA) and a uniform planar array (UPA) , which involves reducing the near-field channel into a combination of multiple far-field channels. A grouped Jensen’s inequality-based column-iteration (GJ-iteration) solution is derived to maximize the SE for the second wideband hybrid precoding subproblem, which addresses the beam split effect in the near-field channel by mitigating the beam split effect in multiple far-field channels. Simulation verification demonstrated that the proposed algorithm enhanced the SE and mitigated the beam split effect.