Miniaturized Antenna Array using Additive Manufactured PLA Substrate with Fractal Geometry and ANN Assisted Fractal Dimensions

This paper address the demand of miniaturized, and rapidly prototyped antenna by utilizing fractal geometry for miniaturization and employing 3D printed substrates for quick prototyping. An approach that integrates artificial neural networks (ANN) into the design process to predict the dimensions of a Sierpinski carpet fractal antenna array developed on a 3D printed polylactic acid (PLA) substrate is presented. The ANN implementation offers a significant advantage by eliminating traditional trial-and-error methods and optimizing the antenna's design dimensions for specific frequency requirements. The PLA substrate, fabricated using the Fused Filament Fabrication (FFF) technique, offers several benefits over other materials, including cost-effectiveness, ease of manufacturing, and the ability to tailor dielectric properties. The dielectric constant and loss tangent of the 3D-printed PLA substrate is calculated using the ring resonator method. Initially, a conventional single rectangle antenna is designed and simulated for a 2.4GHz resonant frequency. Further two iterations of the fractal geometry shift the resonant frequency from 2.4GHz to 1.78 GHz. By employing ANN, we predict the dimensions required to achieve the desired frequency. The design process is conducted in Ansys HFSS environment, where parametric analysis of the second iteration of the fractal geometry is conducted with various patch dimensions. A fractal patch antenna at 1.78 GHz, a 1x2 fractal antenna array at 1.78 GHz, and a 1x2 fractal antenna array at 2.4 GHz resonant frequency are fabricated to validate the design process. The proposed approach achieves a miniaturization of 31.46%, demonstrating good agreement between simulated and measured results. This study showcases the effectiveness of combining fractal geometry with ANN and 3D printing additive manufacturing technology to achieve miniaturized antenna arrays for wireless applications.