Machine Learning Assisted Antenna Design for 5G Applications

The performance of 5G networks is significantly influenced by the design and optimization of antenna. Machine Learning (ML) plays a significant role in antenna design by enabling the optimization of antenna parameters and configurations for enhanced performance. In this paper, the optimal design of an antenna using machine learning, specifically designed to operate at 3.5 GHz is presented. The data acquired from the Parametric Analysis (PA) is employed to develop a regression model. Six different ML regression algorithms like K-Nearest Neighbor (KNN), Random Forest, Decision Tree, XG Boost Regression, Ridge Regression and Support Vector Machine (SVM) are employed. Among these, Decision Tree and Random Forest Regression offer minimum mean square error and better predictions. The antenna designed using both PA and ML are compared and analysed. An antenna design based on ML achieves performance comparable to conventional patch antenna and an optimized antenna design based on PA. Moreover, it exhibits 43% reduction in size when compared to the conventional patch antenna.