Unveiling Novel Approaches: Machine Learning Optimization for Circularly Polarized Antennas Tailored for 5G Communication

This paper advocates the application of Machine learning techniques in the domain of antenna design and optimization to expedite the entire process. The conventional approach to antenna design has been notably time-consuming due to the intricate and computationally intensive numerical methods involved. The utilization of simulation software, such as HFSS, adds to the complexity, providing parameter results based on predefined dimensions or dimension ranges, but at the expense of significant time investment. Consequently, there is a compelling need for optimization to align the antenna’s characteristics with the desired parameters efficiently. In this study, we propose the creation of a circularly polarized dielectric resonator antenna specifically designed for 5G communication and tuned to operate at a resonant frequency below 6 GHz. To enhance the efficiency of the design process, we leverage the capabilities of five distinct machine learning algorithms, namely bagging, boosting, KNN, and ANN. Remarkably, the KNN algorithm, particularly with k=9, demonstrates notable success, achieving an impressive accuracy rate of up to 98.7%. The results obtained through this approach enable the estimation of dimensions for desired parameters, a task previously challenging with HFSS alone. This signifies a significant advancement in antenna design efficiency facilitated by Machine Learning methodologies.