Performance prediction of a two-port terahertz-based graphene–silicon-based tunable antenna with polarization and pattern diversity features using machine-learning algorithms

This paper proposes a hybrid twin-port radiator with polarization and pattern diversity features in the terahertz (THz) frequency domain. Three major contributions of the proposed antenna design are: (i) a rounded wave is obtained by stimulating the silicon ceramic via a fan blade formed slot; (ii) a mirror orientation of aperture and suspension of partially reflecting surface (PRS) helps to get better isolation with an S-component and far-field component, respectively; and (iii) the random forest and XGBoost technique predicts the |<i>S</i> ₁₁| of the designed aerial effectively. A coating of graphene over silicon dielectric makes the radiator tunable. The optimized outcome confirms that the planned aerial operates effectively in between 2.45 and 3.15 THz with less than 3-dB axial ratio from 2.61 to 2.89 THz. PRS tilts the beam gotten from port-1 and port-2 by ±30^∘. The design is suitable for THz-based cellular machines because of all these qualities.