A Reliable Multi-information Entropy Glitch PUF Using Schmitt Trigger Sampling Method for IoT Security

Abstract

Glitch PUF (Physical Unclonable Function) has non-linear characteristics to resist the threat of modeling attacks. However, there are some problems in Glitch PUF such as low information entropy and susceptibility to noise factors. In this paper, a Glitch PUF is proposed which uses multilevel parallel architecture to generate multi-bit stable information entropy. Firstly, the delay time of different paths are controlled by the adjustment module circuit, which changes the glitch width and obtain stable glitch waveforms. A noise reduction circuit is designed to filter the noise by using the hysteresis effect and feedback mechanism of the Schmitt trigger. And then, a pulse width detection circuit is proposed to extract the Glitch PUF output data. Finally, the 128 bits Glitch PUF circuit is implemented in TSMC 65 nm CMOS technology. The experimental results show that the randomness is 99.9% and the uniqueness is 50.03%, which means the proposed design can be widely used for IoT security.