Gain enhancement of SiO2 substrate based fractal antenna integrated with frequency selective surface and its optimization using machine learning algorithms for terahertz utilizations

A terahertz fractal antenna integrated with a metamaterial frequency selective surface (FSS) is designed for gain enhancement purposes. The designed antenna operates in various modes of operation to provide biconical and conical radiation patterns. A novel frequency-selective surface has been constructed and combined with a simple fractal antenna structure to reduce the back radiation from the radiator and improve its gain. Detailed simulation studies have been performed using high-performance 3D EM analysis software. The FSS is also studied using an equivalent electric circuit model approach to verify the operation of the designed radiator system. The prescribed FSS integrated fractal radiator operates over a wide range of terahertz frequency spectrum extending from 3.7 to 10 THz with a relative bandwidth of 91.97%. The design of the FSS structure integrated with a simple patch enhances the gain from 7.78 dBi to 9.28 dBi at 4.06 THz and from 8.71 dBi to 11.9 dBi at 7.27 THz. Additionally, the performance of the proposed FSS integrated fractal aerial has been optimized using machine learning (ML) based techniques. Various machine learning algorithms using XG Boost, Artificial Neural Network, and Random Forest are used to optimize the design parameters and to predict characteristic parameters of the proposed structure at the THz regime. Conventionally, the optimization process by traditional EM solvers like HFSS, CST, etc. requires more processing time in case of designing complex antenna structures. The implementation of a machine learning-based optimization method increases the speed and accuracy of the simulation optimization process and also achieves acceptable performance.