Modeling Attack Resistant Enhanced Cryptographic Key Generation and Optimized Selection Using Crossover Ring Oscillator PUF

With the rapid growth in the use of Internet of Things (IoT) devices, Physically Unclonable Functions (PUFs) have gained prominence as a promising technology for generating unique, device-specific secret keys and bolstering hardware-based security. IoT enables seamless device connectivity and cloud data sharing, making it a prime target for hackers, necessitating encryption with secret keys to ensure data integrity. Although PUFs generate unique secret keys, they remain susceptible to various modeling attacks if they lack sufficient robustness and fail to generate an adequate number of Challenge-Response Pairs (CRPs).In this paper, we propose a robust key generation method using a Crossover Ring Oscillator (CRO) PUF, combined with Transmission Gate Logic (TGL) and a NAND gate-based multiplexer in its intercrossing stages increasing CRPs and measuring under varying environmental conditions (voltage and temperature), our approach makes it significantly harder to identify the responses used for key generation from the device. Additionally, we introduce an algorithm for optimal key selection based on Shannon entropy p-value and Hamming distance criteria, enhancing unpredictability to mitigate vulnerabilities against various modeling attacks. The resilience of these keys is evaluated against several machine learning algorithms, including Logistic Regression (LR), Naive Bayes (NB), Support Vector Machine(SVM), Decision Tree (DT), and K-nearest Neighbor (KNN). Our results demonstrate that the optimally selected secret keys, generated using our algorithm, exhibit strong resistance to these widely used modeling attack techniques.