FPGA-based strong PUF with increased uniqueness and entropy properties

Abstract

Physical unclonable functions (PUFs), are a type of physical security primitive which enable identification and authentication of hardware devices, such as field programmable gate arrays (FPGAs) and application specific integrated circuits (ASICs). Arbiter PUFs were the first proposed Strong PUF and are also widely studied. However, these designs often suffer from poor uniqueness and reliability characteristics leaving them vulnerable to modeling attacks, as well as being difficult to implement on FPGAs due to the physical layout restrictions. Some more recent designs based around non-linear voltage transfer characteristics, or non-linear currents improve the resistance against modeling attacks. However they can only be implemented on ASICs due to their voltage/current requirements. To address this problem, we propose a new PUF circuit that offers a significantly higher theoretical entropy than the traditional Arbiter PUF construction, and which is specifically designed for FPGAs. The proposed work is verified on a low-cost Nexys4 board which contains a Xilinx Artix-7 FPGA fabricated at 28nm. The experimental results give a uniqueness of 20 %, considerably higher than the reported 9 % of a traditional Arbiter PUF design, and an expected reliability of ≈ 96% over an environmental temperature range of 0° C to 75° C, with a reliability of ≈ 92 % with ±10 % variation in supply voltage.