| Title | CT PUF: Configurable Tristate PUF against Machine Learning Attacks |
| Publication Type | Conference Paper |
| Year of Publication | 2020 |
| Authors | Wu, Qiang, Zhang, Jiliang |
| Conference Name | 2020 IEEE International Symposium on Circuits and Systems (ISCAS) |
| Keywords | Hardware, hardware security, Inverters, machine learning, Neural networks, physical unclonable function, Predictive models, security, Training |
| Abstract | Strong physical unclonable function (PUF) is a promising lightweight hardware security primitive for device authentication. However, it is vulnerable to machine learning attacks. This paper demonstrates that even a recently proposed dual-mode PUF is still can be broken. In order to improve the security, this paper proposes a highly flexible machine learning resistant configurable tristate (CT) PUF which utilizes the response generated in the working state of Arbiter PUF to XOR the challenge input and response output of other two working states (ring oscillator (RO) PUF and bitable ring (BR) PUF). The proposed CT PUF is implemented on Xilinx Artix-7 FPGAs and the experiment results show that the modeling accuracy of logistic regression and artificial neural network is reduced to the mid-50%. |
| DOI | 10.1109/ISCAS45731.2020.9180409 |
| Citation Key | wu_ct_2020 |
CT PUF: Configurable Tristate PUF against Machine Learning Attacks