IoT-Enabled PCF-SPR Glucose Sensor With MLP-Based Performance Prediction

This work presents a novel intelligent and connected biomedical sensing platform that integrates a high-sensitivity photonic crystal fiber-surface plasmon resonance (PCF-SPR) sensor with artificial intelligence and Internet of Things (IoT) technologies for real-time glucose monitoring. Finite Element Method (FEM) simulations were performed to investigate the optical response in terms of confinement loss (CL), resonance wavelength shift, and amplitude sensitivity over clinically relevant glucose concentrations. The proposed sensor achieves a high wavelength sensitivity of 2173.91 nm/RIU, amplitude sensitivity of 732.24 RIU^-1, and a detection resolution of 4.6 × 10^-5 RIU. To overcome the computational burden of FEM simulations, a Multilayer Perceptron (MLP) model was developed to accurately predict the full confinement loss, demonstrating excellent agreement with numerical results (R² = 0.9963) and effectively capturing nonlinear parameter interactions. The PCF-SPR sensor is directly interfaced with an ESP32 microcontroller for real-time signal digitization and cloud-based IoT monitoring. This integrated framework enables fast data processing, remote visualization, and intelligent glucose estimation, establishing a seamless transition from optical simulation to smart, connected plasmonic biosensing systems.