Design and Optimization of Cross-Shaped Slot UWB Miniature Patch Antennas for 28 GHz 5G mmWave Applications Using ANN Based on k-Fold Cross-Validation

This paper introduces an innovative method for designing and optimizing an Ultra-Wideband (UWB) patch antenna utilizing an Artificial Neural Network (ANN). The main objective is to accurately predict the essential dimensions of the patch antenna to meet specific resonance and bandwidth requirements, thereby streamlining the design process. The ANN model is rigorously trained and validated using the k-fold cross-validation method, ensuring robust performance and accurate predictions across multiple data subsets. The results demonstrate that the ANN model can precisely forecast the dimensions of the patch antenna, achieving a resonance frequency of 28 GHz and a bandwidth of 20 GHz. The proposed antenna exhibits a maximum gain of 5.75 dB and an efficiency of 99%, which is on par or better than existing designs. The effectiveness of the proposed model is further validated through simulations using HFSS and CST software, showing excellent agreement in reflection coefficient, gain, and efficiency.