Energy-Efficient Target-Aware Hybrid Beamforming for THz Near-Field ISAC with Sparse Connectivity

Nusaibah A. Alshorman, Chong Han, Huseyin Arslan

Integrated sensing and communication (ISAC) at terahertz (THz) frequencies enables ultra-high-resolution perception while facing a key limitation: highly directional THz beams cannot illuminate extended targets within a single beam. Conventional solutions rely on sequential beam scanning, reducing sensing accuracy and increasing energy consumption. Moreover, in conventional sparse-array, grating lobes are generally treated as undesirable artifacts that should be suppressed to avoid ambiguity and interference. In contrast, this paper adopts a reverse design philosophy by intentionally engineering sparsity-induced grating lobes as controllable auxiliary illumination beams for extended-target sensing. This paper exploits grating lobes and proposes a sparse-connected hybrid beamforming architecture that intentionally engineers and exploits grating lobes to enable single-shot, full-aperture illumination of extended targets while supporting multi-user downlink communication. A switch-controlled sparse RF network preserves the array aperture and generates a dominant main lobe with structured secondary lobes covering the entire target extent. A covariance-driven alternating-minimization framework jointly optimizes digital precoders, quantized phase shifters, and antenna-RF switching. Simulations at 140 GHz demonstrate near fully-digital Cramer-Rao sensing accuracy, competitive communication performance in low-rank THz channels, rapid convergence, and significant hardware and energy savings, establishing structured sparse connectivity as a scalable and energy-efficient solution for extended-target THz ISAC.