AI/ML-Based Design and Analysis of a Four-PIN-Diode Frequency-Reconfigurable Multiband Antenna with Slot and Defected Ground Structure

Abstract

In this paper, an AI/ML aided frequency reconfigurable multiband antenna for wireless application in S and C band is presented. The proposed antenna utilizes four PIN diodes, slot loading and defected ground structure (DGS) to alter the surface-current distribution for multiple operating frequencies in the range of 3–8 GHz. The ANSYS HFSS model is used to generate an electromagnetic dataset that changes the antenna geometrical parameters and the switching configuration of the PIN-diodes. Important antenna parameters such as resonant frequency, S₁₁, bandwidth and gain are predicted by machine-learning models like Random Forest, Support Vector Regression, Decision Tree and Artificial Neural Network. The optimum performed ML model is then employed as a surrogate model to minimize the number of full-wave simulations for optimizing the antenna during antenna optimization. The optimized design is printed on an FR-4 substrate and tested in experimental circuit using a vector network analyzer. The proposed AI/ML-based system offers a systematic and computationally efficient approach to design multi-state frequency-reconfigurable antennas for 5G/6G, IoT, radar, satellite and other wireless applications.