Paper

A Dynamic Soft Decision Fuzzy Extractor for a CMOS Image Sensor PUF

2018 Nov27 -Nov30

Authors Shunsuke Okura*, Ryota Ishiki**, Masayoshi Shirahata**, Takaya Kubota**, Mitsuru Shiozaki**, Kenichiro Ishikawa*, Isao Takayanagi*, and Takeshi Fujino**

*Brillnics Japan Inc.,
**Research Organization of Science and Engineering Ritsumeikan University

ISPACS2018

It is required to ensure the sensor device security that is an information interface at IoT age. For imaging devices, a CMOS image sensor (CIS) physically unclonable function (PUF) was proposed for device authentication. However, any bit error in the PUF response is not acceptable for cryptographic purpose. Conventionally, a soft decision fuzzy extractor was presented to regenerate a encryption key repeatably in which error is corrected with a helper data. The helper data is a reliability information of each PUF response bit and requires large circuit overhead. In this paper, a novel dynamic soft decision fuzzy extractor is proposed for the CIS PUF, which utilizes Gaussian error function to dynamically derive the reliability of each PUF response bit. High error correction capability with small circuit overhead of the proposed fuzzy extractor is confirmed with measurement and simulation results. Index Terms—IoT, hardware security,