SpongePUF: A Modeling Attack Resilient Strong PUF with Scalable Challenge Response Pair
Zhenzhe Chen, Takashi G. Sato, Hirofumi Shinohara · 2024
Traditional physically unclonable functions (PUFs) have difficulty in satisfying the varying challenge-response pair (CRP) length requirements across different scenarios. This paper presents a strong PUF that supports scalable response lengths. Utilizing the Sponge structure, a 100-bit challenge input can generate a CRP with a response length ranging from 20 to 100 bits. Using a non-linear block with bent functions, the generated responses of varying lengths pass NIST SP 800–22 and other statistical tests. The PUF exhibits the highest resilience to modeling attacks by 20M training CRPs and the highest 5 bits/cycle throughput using the 100-bit response mode compared with those of other works.