Frequency reconfigurable wideband rectangular dielectric resonator antenna for sub-6 GHz applications with machine learning optimization

This work presents frequency reconfigurable Wideband Rectangular Dielectric Resonator Antenna (RDRA) using a Machine Learning (ML) approach for 5G (Sub-6 GHz) applications. The concept of ML is integrated with reconfigurable DRA for the first time. A novel approach has been proposed to obtain wideband and frequency reconfigurability. Frequency reconfigurability is achieved electronically (PIN diode switches) and operates in four different configurations, offering a maximum of 76.84 % wide tuning range. Hybrid structure and higher order mode TE111 and TE211 modes excitation is responsible for wideband operation. The Extreme Gradient Boosting (XGB) ML algorithm provides more than 90 % accuracy in all configurations for S11 prediction.

Automated summary

This work presents a frequency reconfigurable wideband rectangular dielectric resonator antenna (RDRA) for 5G applications using a machine learning approach. The concept of machine learning is integrated with reconfigurable DRA for the first time. A novel approach is proposed to achieve wideband and frequency reconfigurability. Frequency reconfigurability is electronically achieved using PIN diode switches and can operate in four different configurations, offering a maximum tuning range of 76.84%. The Extreme Gradient Boosting (XGB) machine learning algorithm provides over 90% accuracy in predicting S11 in all configurations.