Design and optimization of circularly polarized dielectric resonator-based MIMO antenna using machine learning for 5G Sub-6 GHz

In this paper, a two-port dielectric resonator (DR) based multiple-input multiple-output (MIMO) antenna is designed and analyzed for the fifth generation (5G) region under the sub-6 GHz band. The cylindrical-shaped ceramic is excited by using a tilted L-shaped aperture. This feeding structure generates the circularly polarized (CP) waves and HEM11δ mode inside the ceramic. The proposed antenna covers the potential 5G band ranging from 2.8 to 3.05 GHz having left-handed circular polarization characteristics. The proposed antenna is used to create the datasets by performing parametric analysis in HFSS software. The datasets are used to optimize the reflection coefficient (S11), isolation (S21), and axial ratio (AR) of the proposed antenna design with the help of several machine learning algorithms. The machine learning algorithms used in the proposed work are Decision Tree (DT), Deep Neural Network (DNN), k-Nearest Neighbors (KNN), Random Forest (RF), and Extreme Gradient Boosting (XGBoost). The results obtained by these algorithms are in line with simulated values of HFSS software. Nevertheless, results obtained by DNN, KNN and RF are significantly better than DT and XGBoost.