Correlation Integral-Based Intrinsic Dimension: A Deep-Learning-Assisted Empirical Metric to Estimate the Robustness of Physically Unclonable Functions to Modeling Attacks

Security Analysis of Digital-Based Physically Unclonable Functions: Dataset Generation, Machine Learning Modeling, and Correlation Analysis

“…The behavior of the PUFs can be effectively represented through a set of collected CRPs [7]. This modeling approach does not require any auxiliary information and solely relies on computations performed on the CRPs themselves [8]. As a consequence, PUFs are susceptible to ML-based attacks [9], especially when CRPs are accessible outside the chip without any protection mechanisms in place.…”

“…The efficiency and simplicity of ML models present a significant threat to PUFs, making them vulnerable to ML-based modeling attacks. In the literature, various ML algorithms have been employed to predict PUF responses, including Decision Tree (DT), Random Forest (RF), Logistic Regression (LR), K Nearest Neighbor (KNN), Support Vector Machine (SVM), kernel-based SVM, Evolutionary Strategies (ES), and Neural Network (NNet), with LR, SVM, and NNet showing dominant performance [8]- [10].…”

Boosted PUF: Boosting Efficiency and Resilience in Configurable RO PUF for IoT Devices

“…Also, the distributed nature of the semiconductor industry with multiple vendors requires a secure mechanism for authentication that relies on the hardware [2]. Physical Unclonable Functions (PUFs) are promising solutions as they support low-cost authentication protocols [3]- [5] and can protect ICs against counterfeiting [6]. Silicon PUF (SPUF) leverages the variation in semiconductor manufacturing to extract device signatures [7].…”