Design and Optimization of a Compact Bio-Inspired Microstrip Antenna at 2.4 GHz for WBAN Applications with Machine Learning based S11 Prediction

This study introduces a novel bio-inspired microstrip patch antenna operating at 2.4 GHz, optimized for Wireless Body Area Network (WBAN) applications through integration with machine learning-based analysis. The antenna features a papaya leaf-inspired radiating element and incorporates two circular slots within a defected ground plane to enhance impedance matching and achieve substantial miniaturization. Fabricated on an FR-4 substrate (dielectric constant: 4.4, loss tangent: 0.02), the device achieves a compact footprint of 53.32 × 32.13 × 1.58 mm³. The leaf-shaped design maximizes effective perimeter while minimizing the physical size, leading to improved radiation efficiency and performance in constrained environments. To accelerate performance evaluation and design iteration, a Random Forest Regression algorithm is implemented for predictive modeling of the S11 reflection coefficient. The model achieves a high degree of accuracy (R² = 0.9015) in mapping the complex behavior of reflection loss, demonstrating its efficacy for ISM band services such as Wi-Fi, WiMAX, and Bluetooth communications. Comparative analysis between simulation and experimental results is conducted, including investigations of substrate variation (PDMS) and impact of slot geometry on antenna performance. The prototype is further integrated with a transceiver module and validated for real-time WBAN scenarios. By embedding a data-driven approach into the design workflow, reliance on computationally expensive electromagnetic simulations is reduced, enabling rapid optimization and deployment of advanced antenna systems for biomedical and wearables applications.