A Modeling Attack on the Sub-threshold Current Array PUF
Yun Liu, Yongliang Chen, Xiaole Cui · 2022
Physical unclonable function (PUF) is regarded as the root of trust of integrated circuits. However, the modeling attack poses a great threat to the PUF circuits. Recently, the non-linear characteristic in the sub-threshold region of MOSFET transistors was applied to construct the PUF circuit against the modeling attacks. This work attacks the sub-threshold current array PUF (SCA-PUF) using an artificial neural network (ANN) model. The ANN model is deliberately designed based on the working principle of the target PUF. In the attacks, the linear functions are used to approximate the nonlinear characteristics of the sub-threshold region, by using the first-order Taylor expansion. The attack results show that the prediction accuracy of the ANN based method reaches 94.4% with 1000 training CRPs, whereas the prediction accuracy of LR algorithm and SVM algorithm are only 75.2% and 77.8%, respectively. It shows that the SCA-PUF is also vulnerable to the modeling attacks, and the ANN method is more powerful compared with the LR and SVM algorithms for the attacks of SCA-PUF.