A chaotic strong transition effect ring oscillator PUF with effective immunity to modeling attacks

Abstract

Although the rapid development of information technology has brought unprecedented convenience and revolutionary changes to human lifestyles, it has also led to serious privacy and security issues. Massive amounts of sensitive information are exchanged within cyber-physical systems (CPS), and the storage and authentication processes of this information are highly susceptible to the attention of malicious attackers. Strong physical unclonable function (PUF) is a crucial hardware security primitive for identity authentication in lightweight internet of things (IoT) devices. However, most of the existing strong PUFs have failed to resist advanced modeling attacks based on machine learning. The paper proposes a chaotic strong transition effect ring oscillator (CS-TERO) PUF, which combines the nonlinearity of strong TERO and the unpredictability of chaotic design to resist modeling attacks. The strong PUF feature is achieved through configurable dual XOR gates and extended TERO chains. Chaotic operations are performed by obfuscating challenges using logical mapping functions. Experimental results demonstrate that the CS-TERO PUF successfully resists state-of-the-art modeling attacks, and the accuracy of all attacks is below 70 %. Compared to other modeling attack-resilient PUFs, our CS-TERO PUF requires fewer hardware resources, saving almost 25 % of area overhead compared to the most lightweight solutions. Additionally, the CS-TERO PUF demonstrates a reliability of 98.27 %, uniqueness of 53.65 %, and uniformity of 49.26 %, reaching an advanced strong PUF application standard. Furthermore, the randomness has also passed the internationally accredited NIST 800-22 S P test.