Performance prediction of dielectric resonator based MIMO antenna for sub-6.0 GHz using machine learning algorithms

Abstract

ABSTRACTIn this article, a dual port dielectric resonator antenna is modeled using various machine learning algorithms i.e. deep neural network (DNN), Random Forest, and XG boost. The unique properties of proposed article are as follows: (i) two different diversity techniques i.e. pattern (with the help of metallic wall) and polarization (mirror image of the aperture) improves the isolation value between the ports; (ii) machine learning (ML) algorithms are used to optimize and predict the reflection coefficient as well as mutual coupling of proposed antenna. The accuracy of ML algorithms is verified by using HFSS EM simulator as well as experimental validation. Error is less than 1–2% between the value predicted from ML algorithms and HFSS/Experimental results. The proposed design is working well in between 2.4–4.02 GHz with 3-dB axial ratio from 2.84–2.95 GHz. All these features make the radiator employable to sub-6.0 GHz frequency band.