Physics-Guided Surrogate-Assisted Closed-Loop Electromagnetic Optimization for Simulation-Efficient IoT Antenna Synthesis

This dataset supports the manuscript titled: "Physics-Guided Surrogate-Assisted Closed-Loop Electromagnetic Optimization for Simulation-Efficient IoT Antenna Synthesis" The dataset was generated using CST Studio Suite through parametric full-wave electromagnetic simulations of a rectangular microstrip patch antenna fabricated on an FR4 substrate (εr = 4.3, thickness = 1.5 mm). A total of 56 valid simulations were conducted by varying: Patch length (L): 26–33 mm Patch width (W): 34–41 mm For each simulation, the following parameters were extracted: Resonance frequency (GHz) Reflection coefficient S11 (dB) This dataset was used to train a Random Forest regression model and to enable closed-loop optimization using differential evolution. The dataset reflects the nonlinear relationship between antenna geometry and electromagnetic performance and supports reproducibility of the results presented in the manuscript. File included: final_production_data.csv (primary dataset used for training and validation) This dataset is publicly available to support transparency, reproducibility, and future research in machine learning-assisted antenna design.