Lightweight obfuscation techniques for modeling attacks resistant PUFs

Abstract

Building lightweight security for low-cost pervasive devices is a major challenge considering the design requirements of a small footprint and low power consumption. Physical Unclonable Functions (PUFs) have emerged as a promising technology to provide a low-cost authentication for such devices. By exploiting intrinsic manufacturing process variations, PUFs are able to generate unique and apparently random chip identifiers. Strong-PUFs represent a variant of PUFs that have been suggested for lightweight authentication applications. Unfortunately, many of the Strong-PUFs have been shown to be susceptible to modelling attacks (i.e., using machine learning techniques) in which an adversary has access to challenge and response pairs. In this study, we propose an obfuscation technique during postprocessing of Strong-PUF responses to increase the resilience against machine learning attacks. We conduct machine learning experiments using Support Vector Machines and Artificial Neural Networks on two Strong-PUFs: a 32-bit Arbiter-PUF and a 2-XOR 32-bit Arbiter-PUF. The predictability of the 32-bit Arbiter-PUF is reduced to ≈ 70% by using an obfuscation technique. Combining the obfuscation technique with 2-XOR 32-bit Arbiter-PUF helps to reduce the predictability to ≈ 64%. More reduction in predictability has been observed in an XOR Arbiter-PUF because this PUF architecture has a good uniformity. The area overhead with an obfuscation technique consumes only 788 and 1080 gate equivalents for the 32-bit Arbiter-PUF and 2-XOR 32bit Arbiter-PUF, respectively.

Machine Learning Attacks and Countermeasures for PUF-Based IoT Edge Node Security

“…Hence, higher non-linearity in a given PUF architecture can improve the uniqueness and randomness with increased defense against modeling attack. Other approaches to ML-resistant PUFs have been randomized challenges [42], obfuscation [12,23], and sub-string-based challenges [28]. Rostami et al presented a prover-verifier framework for successful authentication based on a subset of response substring [28].…”

“…We also evaluate the performance of the cloning models when the challenge is obfuscated, as postulated in obfuscated PUF architectures such as [12,23]. We consider a simpler version of these approaches for our experiments.…”

“…It should be noted that the performance is tested only on a limited evaluation set of 200 CRPs. More complicated obfuscation techniques such as those proposed in [12,23] and less training would further degrade the performance of machine learning-based cloning attacks.…”

“…Inspired from the pioneering work of Goodfellow et al [13] on Generative Adversarial Networks (GANs), we propose a machine learning-based defense, a discriminator, to identify the possibility of cloning using any ML-based attack non-invasive attack. Extant countermeasures [23,28] to MLbased cloning have focused on creating complex cloningresistant PUF architecture. As we enter into a more realizable IoT ecosystem, complex PUF architectures may not be suitable for lightweight IoT systems.…”

Machine-Learning Attacks on PolyPUFs, OB-PUFs, RPUFs, LHS-PUFs, and PUF–FSMs

“…Nevertheless, the authors of all five protocols deserve credit for their compliance with Kerckhoffs' principle [27], i.e., the specification of the obfuscation logic is public. On the contrary, Mispan, Su, Zwolinski, and Halak [28], [29] recently proposed three PUF-based authentication protocols of which the specification is private, i.e., they rely on the almost universally rejected paradigm of security through obscurity. To be clear: obscurity is the only protective measure, which differs from cases where obscurity augments an inherently secure system.…”