Novel RPM technique to dismiss systematic variation for RO PUF on FPGA

Abstract

Physical Unclonable Function (PUF) is a function that cannot be modeled as it utilizes the random process variations on a silicon chip to generate a unique bit stream of `1's and `0's (response bits) which can be used for authentication and cryptography applications. As PUF is highly rely upon process variations, the response bits generated are governed by the systematic process variation instead of the stochastic process variation, which will reduce the randomness in the response bits. In this paper we propose the novel Random Patch Mixer (RPM) technique to dismiss the systematic variation effect on the response bits generated. We applied the RPM technique on data obtained from 29 Spartan 3E FPGA chips. We showed that our RPM technique has successfully dismissed the systematic variation effect on the response bits generated from the ROPUF on FPGA. We also proved that the responses generated by applying the RPM Technique passed the National Institute of Standards and Technology NIST statistical test for randomness.