Secure and Energy-Efficient FPGA Smart Sensor for Real-Time Nanocomposite Strain Sensing with Deep Neural Network and AES

Ensuring the structural integrity of safety-critical systems, from aerospace to civil infrastructure, demands continuous, real-time strain monitoring under stringent security and power constraints. The integration of cryptographic circuits with Internet of things connectivity and nanostructured sensors offers a promising path toward enhancing system safety. However, challenges including noise, hysteresis, signal reproducibility, modeling, calibration, and data security risks during wireless transmission make signal processing difficult. This study presents a secure, energy-efficient smart strain sensing system for piezoresistive nanocomposite films, implemented on a field-programmable gate array (FPGA). The system integrates a deep neural network model to approximate the average piezoresistive response and mitigate static nonlinearities, alongside a 128-bit Advanced Encryption Standard (AES) cryptographic engine to secure Bluetooth data transmissions. This hardware-accelerated approach enables continuous structural strain monitoring, providing reliable data for early failure warning and predictive maintenance models. Experimental results demonstrate that the secure smart strain sensing system integrated with AES-encryption achieves a bit-error rate of 1.2×10−7, a packet delivery ratio of 99.7%, and a mean square error of 2.4×10−6, confirming the fidelity of the encrypted Bluetooth transmission under the tested conditions (0 dBm, 10 m range). In addition, a 99.4% dice similarity coefficient is achieved between transmitted and received/decrypted data, with a normalized mean absolute error of 12.5% over the elastic range. The system achieves a simulated core dynamic power consumption of 0.74 mW while utilizing only 8936 logic cells and 953 registers on an Altera 5CSEMA5F31C6N FPGA. Compared to existing approaches, the proposed system offers significant advantages in security, energy efficiency, and hardware resource utilization, making it a promising solution for real-time embedded monitoring in reliability-critical applications.