Data-driven modeling of Si–graphene nano optical antenna with high-isolation for 6G communication system
摘要
This article presents the design and analysis of a compact two-port Si–graphene-based dielectric resonator antenna operating in THz band for 6G applications. Circular polarization is achieved through diagonal perturbation of a cylindrical silicon dielectric resonator, while frequency tunability is enabled by coating the resonator with a monolayer graphene sheet. A key feature of the proposed configuration is the reverse placement of the perturbed dielectric resonators, which generates opposite circular polarization senses at the two ports, resulting in polarization-diversity-driven isolation exceeding 25 dB. The antenna operates over an impedance bandwidth of 0.75 THz (2.2–2.95 THz), with a corresponding 3-dB axial-ratio bandwidth of 0.35 THz (2.3–2.65 THz), and exhibits a moderate realized gain of approximately 6.5 dBi. To reduce the computational burden associated with full-wave simulations in the terahertz regime, machine-learning-based surrogate models using Random Forest and XGBoost regressors are employed. The proposed work highlights a physics-consistent integration of circular polarization, tunability, polarization-diversity-based isolation, and ML-assisted analysis within a compact terahertz MIMO antenna framework.