Master of Engineering Science
Electrical and Computer Engineering
A Physically Unclonable Function (PUF) is a new and promising approach to provide security for physical systems and to address the problems associated with traditional approaches. One of the most important performance metrics of a PUF is the randomness of its generated response, which is presented via uniqueness, uniformity, and bit-aliasing. In this study, we implement three known PUF schemes on an FPGA platform, namely SR Latch PUF, Basic RO PUF, and Anderson PUF. We then perform a thorough statistical analysis on their performance. In addition, we propose the idea of the Hybrid PUF structure in which two (or more) sources of randomness are combined in a way to improve randomness. We investigate two methods in combining the sources of randomness and we show that the second one improves the randomness of the response, significantly. For example, in the case of combining the Basic RO PUF and the Anderson PUF, the Hybrid PUF uniqueness is increased nearly 8%, without any pre-processing or post-processing tasks required. Two main categories of applications for PUFs have been introduced and analyzed: authentication and secret key generation. In this study, we introduce another important application for PUFs. In fact, we develop a secret sharing scheme using a PUF to increase the information rate and provide cheater detection capability for the system. We show that, using the proposed method, the information rate of the secret sharing scheme will improve significantly.
Khoshroo, Sasan, "Design and Evaluation of FPGA-based Hybrid Physically Unclonable Functions" (2013). Electronic Thesis and Dissertation Repository. 1281.