AI modelling-based reconfigurable dual band antenna for GPS and ISM bands with metamaterial superstrate for high gain application

A reconfigurable dual-band antenna operating at L band and ISM band is introduced in this paper, suitable for GPS and WiFi applications. The antenna has a primary and secondary radiating patch, which resonates at two different frequencies depending on the reconfigurability. A split-ring resonator (SRR) based metamaterial is designed and etched on the primary patch antenna, enabling an optimum solution of reconfigurability through a diode-based switching circuit that controls the performance with respect to single or dual band operation with its ON and OFF states. The novelty of the paper lies in embedding a diode in SRR, which allows the geometry transition from a traditional split ring to a concentric ring configuration with bias control. This shape change significantly alters the current distribution and surface wave interaction, enabling precise control over the antenna’s resonant characteristics. A three-diode switching network is used to independently and synchronously manage the radiating patch structure and the SRR metamaterial. Additionally, a passive metamaterial array of circular patches is employed as a superstrate to meet the high gain requirements. This proposed antenna was fabricated and tested for performance evaluation of impedance matching and gain across both bands. There is a significant 4dB increase in gain with the superstrate layer, offering a practical solution for advanced wireless communication systems. The paper also presents the analysis of the dual-band antenna dataset using Artificial Intelligence (AI) models. The dataset comprises three antenna parameters of return loss, VSWR, and gain, which are used to train and evaluate using a Convolutional Neural Network (CNN) and a Support Vector Machine (SVM). The collected data is systematically divided into training, testing, and validation sets to ensure robust model performance. The CNN and SVM models achieved a training accuracy of 99.28% and 95% and testing accuracy of 95% and 98% respectively. The results demonstrate effectiveness in accurately classifying antenna parameters. It highlights the potential of machine learning in analyzing antenna characteristics in wireless communication.