Machine Learning Enabled Al<sub>2</sub>O<sub>3</sub>Ceramic Based Dual Band Frequency Reconfigurable Dielectric Antenna for Wireless Application
Jayant Kumar, Pinku Ranjan, Rakesh Chowdhury
Abstract
A ceramic ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Al</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">O</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> ) material based dual-band high-tuning range frequency reconfigurable dielectric antenna for wireless applications with Machine Learning (ML) algorithm is presented in this article. The proposed antenna is a hybrid structure in which the antenna radiator is designed with a Dielectric Resonator (DR) (Alumina ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Al<sub>2</sub></i> O <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> ) ceramic material with a relative dielectric constant (∈ <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><sub>r</sub></i> )=9.8. The presented work offers dual-band, compactness, and frequency reconfigurability (FR).FR is obtained through two PIN diode switches, operating in ON-ON, ON-OFF, OFF-ON and OFF-OFF configurations. It offers a total spectrum and a maximum wide tuning range of 71.49 % and 44.44 %, respectively. Dual-band is generated through the excitation of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">HEM</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">11δ</sub> , and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">HEM</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">12δ</sub> mode in cylindrical Dielectric Resonator (CDR). In contrast, compactness is obtained through the higher-order mode excitation and hybrid structure. The proposed antenna is designed on the ANSYS HFSS software and optimized through various ML algorithms such as K-Nearest Neighbor (KNN), Artificial Neural Network (ANN), Decision Tree (DT), Extreme Gradient Boosting (XGB), and Random Forest (RF). In all configurations, KNN achieved more than 99 % accuracy for the prediction of reflection coefficient ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">s</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">11</sub> ). The proposed antenna is used for WiMAX, WLAN, and 5G wireless applications.