A physics-informed Airy beam learning framework for blockage avoidance in sub-terahertz wireless networks

The line-of-sight blockage is one of the main challenges in sub-terahertz wireless networks. Interestingly, the extended near-field range of sub-terahertz nodes gives rise to near-field wavefront shaping as a feasible remedy to tackle this issue. Recently, Airy beams emerged as one promising solution that opens significant opportunities to circumvent blockers with unique self-accelerating properties and curved trajectories. Yet, to unleash the full potential of curved beams in practice, one fundamental challenge remains: How to find the best beam trajectory? In principle, an infinite number of trajectories can be engineered. To find the optimal trajectory, we develop a physics-informed machine-learning framework for Airy beam shaping based on a detailed understanding of near-field electromagnetics, ray optics, and wave optics. The experimental results indicate that Airy beams, when correctly configured, can substantially increase the link budget under high-blockage scenarios even compared to near-field beam focusing, providing insight into coverage expansion and blind-spot reduction.