Dual-Band Fractal Antenna for 5G and ISM Band Applications with ML-based Performance Forecasting

This article presents the design of a dual-band hybrid fractal antenna. The research presents a combination of simulation and experimental verification along with an equivalent circuit model to evaluate the antenna's performance . The hybrid fractal antenna is an amalgamation of Vicsekcross and Sierpinski carpet geometries, so it is termed the Vicsekcross Sierpinski Antenna (VSA). The proposed antenna has a coaxial feed positioned at 45o with a dielectric substrate RT/Duroid 5880TM (=2.2). A maximum realized gain response of 7.54 and 5.87 dBi is observed in the frequency ranges of 3.55–3.65 & 5.69–5.92 GHz. The radiator operational frequency bands are 0.10 GHz (3.55–3.65 GHz) and 0.23 GHz (5.69–5.92 GHz). The impedance bandwidth for both operational bands is 2.77% & 3.96%. The hybrid antenna has a dimension of 0.83λ0×83λ0×0.038λ0 at 3.6 GHz and 1.34λ0×1.34λ0×0.062λ0 at 5.8 GHz. Furthermore, a supervised regression machine learning (ML) algorithm is adapted for antenna’s resonant frequency and realized gain prediction. Random Forest Regression has highest accuracy among others, with R-squared values of 93.54% for reflection coefficient and 91.48% for gain, confirming the effectiveness of the ML-assisted design for ISM band and 5G applications.