A novel offset method for improving bitstring quality of a Hardware-Embedded delay PUF

Abstract

Statistical properties including uniqueness, randomness and reproducibility are commonly used as metrics for Physical Unclonable Functions (PUFs). When PUFs are used in authentication protocols, the first two metrics are critically important to the overall security of the system. Authentication reveals the bitstrings (and helper data if used) to the an adversary, and makes the PUF vulnerable to tactics that can lead to successful cloning and impersonation. In this paper, we investigate security metrics including Entropy, uniqueness and randomness using hardware data collected from a set of 45 Xilinx Zynq FPGAs which implements a Hardware-Embedded Delay PUF called HELP. HELP measures and analyzes variations in path delays that occur within a hardware-implemented macro. A novel technique is proposed that allows the verifier to randomly or purposefully offset path delays to obfuscate (in the former case) and/or tune (in the latter case) the bitstring generation process. We show that tuning additionally has a significant impact on the statistical quality of the bitstrings.