NIST-Lite: Randomness Testing of RNGs on an Energy-Constrained Platform

Lee, Cheng-Yen; Bharathi, Kunal; Lansford, Joellen; Khatri, Sunil P.

doi:10.1109/iccd53106.2021.00019

Cited by 4 publications

(1 citation statement)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Randomness is adopted to describe the unpredictability of the bit sequence generated by the PUF. Ideally, a PUF should output zero and unity with an equal probability (50%) [24]. Based on the NIST SP 800-22 randomness test suite, we Additionally, we employed a cross-correlation function (CCF) to assess the correlation between the rising and falling edges.…”

Section: Randomness and Correlationmentioning

confidence: 99%

PN-APUF: A MOSFET threshold loss based strong arbiter PUF with double-edge sampling

Shao,

Ma,

et al. 2024

IEICE Electron. Express

View full text Add to dashboard Cite

Arbiter physically unclonable function (APUF) is one of the most representative strong PUFs. Improving traditional APUF can enhance the performance of APUF-based PUFs. Accordingly, based on the theoretical foundation of traditional APUF, this paper proposes a novel PUF that utilizes the threshold loss, in which the PMOS transmits at a low level and the NMOS transmits at a high level, as the deviation source. In this manner, we significantly extended the delay deviation of both the rising and falling edges, thereby improving the randomness of PUF. The proposed PUF passed six subtests of the NIST randomness test suite. The delay unit only uses 20 minimum-size transistors with a fully custom layout area of 6.14 μm 2 in the TSMC 65-nm process, reducing the area by 15% compared to traditional APUF. The core area feature size is 93,018 F 2 which is more than sixfold reduction to the voltage-based PUFs. Furthermore, the proposed PUF generates twice the number of challenge-response pairs by sampling the rising and falling edges. It also exhibits a superior resistance against machine learning attacks compared to APUF, making it a better alternative.

show abstract

Section: Randomness and Correlationmentioning

confidence: 99%