Design Of Secure Physical Unclonable Function

Tee, Kah Keat (2019) Design Of Secure Physical Unclonable Function. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (24 Pages) Design Of Secure Physical Unclonable Function.pdf - Submitted Version Download (528kB)

Abstract

Physical Unclonable Functions (PUFs) are new primitive security based on hardware. It uses the random physical disorder or process variations to deliver specific input challenges response, called the Challenge-Response Pairs (CRP). PUF can classify into two categories the Weak PUF and the Strong PUF. Strong PUFs have many potential challenges and weak PUF normally has only a few numbers of CRPs. Strong PUFs are unpredictable and cheap authentication. Arbiter PUF is one of the kinds of strong PUF. But because of its deterministic logic, the modeling techniques can break it in a short time. Therefore the objective of this project is to develop a soft model of Arbiter PUF after that proposed a technique to increase the resistance of strong PUF against modeling attacks. The effectiveness of the proposed technique is using artificial neural network machine learning algorithm to test it. The resulting outcome is measure the predictivity, uniformity, and uniqueness. The result shows that the resistance of arbiter PUF again to the modeling attack of ML is low, so the adversary is easy to exploiting model-building attacks to break it. But, if adding a strong PUF and XOR gate the statistical analysis shows a great increase in the resistance against the modeling attacks. Therefore by using this method, the strong PUF able to reduce the predictability of PUF response to below 60% as the expectable result.

Actions (login required)

View Item