A High-Reliability Gas-Sensor-Based PUF for Secure IoT

This work introduces a Physical Unclonable Function (PUF) architecture based on WO2.90gas sensors, generating cryptographic keys directly from the microscopic structural randomness of nanomaterials. We propose a multi-dimensional resistive arbitration algorithm to dynamically map the sensors' inherent resistive variations into a scalable challenge-response space. Systematic experimental evaluations demonstrate an ideal uniqueness of 50.02% and an exceptionally low average Bit Error Rate (BER) of 1.45%. Furthermore, the PUF maintains high operational robustness across varying voltages, temperatures, and time drift. To validate its cryptographic rigor, a comprehensive dataset of 10,000 challenge-response pairs (CRPs) was evaluated. The results confirm that the generated bitstreams successfully pass the NIST SP 800-22 statistical test suite, yielding an effective min-entropy of 127.84 bits for 128-bit responses. Additionally, the architecture exhibits strong resilience against classic machine learning modeling attacks. These physical and cryptographic validations establish the proposed gas-sensor PUF as a highly secure, scalable, and reliable hardware primitive.