We show in this paper how several proposed Physical Unclonable Functions (PUFs) can be broken by numerical modeling attacks. Given a set of challenge-response pairs (CRPs) of a PUF, our attacks construct a computer algorithm which behaves indistinguishably from the original PUF on almost all CRPs. This algorithm can subsequently impersonate the PUF, and can be cloned and distributed arbitrarily. This breaks the security of essentially all applications and protocols that are based on the respective PUF. The PUFs we attacked successfully include standard Arbited PUFs and Ring Oscillator PUFs of arbitrary sizes, and XO Arbiter PUFs, Lightweight Secure PUFs, and Feed-Forward Arbiter PUFs of up to a given size and complexity. Our attacks are based upon various machine learning techniques including Logistic Regression and Evolution Strategies. Our work leads to new design requirements for secure electrical PUFs, and will be useful to PUF designers and attackers alike.

/pdf/modeling-attacks-on-physical-unclonable-functions-uqecmaqblx.pdf

Modeling attacks on physical unclonable functions

We discuss numerical modeling attacks on several proposed strong physical unclonable functions (PUFs). Given a set of challenge-response pairs (CRPs) of a Strong PUF, the goal of our attacks is to construct a computer algorithm which behaves indistinguishably from the original PUF on almost all CRPs. If successful, this algorithm can subsequently impersonate the Strong PUF, and can be cloned and distributed arbitrarily. It breaks the security of any applications that rest on the Strong PUF's unpredictability and physical unclonability. Our method is less relevant for other PUF types such as Weak PUFs. The Strong PUFs that we could attack successfully include standard Arbiter PUFs of essentially arbitrary sizes, and XOR Arbiter PUFs, Lightweight Secure PUFs, and Feed-Forward Arbiter PUFs up to certain sizes and complexities. We also investigate the hardness of certain Ring Oscillator PUF architectures in typical Strong PUF applications. Our attacks are based upon various machine learning techniques, including a specially tailored variant of logistic regression and evolution strategies. Our results are mostly obtained on CRPs from numerical simulations that use established digital models of the respective PUFs. For a subset of the considered PUFs-namely standard Arbiter PUFs and XOR Arbiter PUFs-we also lead proofs of concept on silicon data from both FPGAs and ASICs. Over four million silicon CRPs are used in this process. The performance on silicon CRPs is very close to simulated CRPs, confirming a conjecture from earlier versions of this work. Our findings lead to new design requirements for secure electrical Strong PUFs, and will be useful to PUF designers and attackers alike.

http://sharps.org/wp-content/uploads/RUHRMAIR-TIFS13.pdf

PUF Modeling Attacks on Simulated and Silicon Data

We discuss numerical modeling attacks on several proposed Strong Physical Unclonable Functions (PUFs). Given a set of challenge-response pairs (CRPs) of a Strong PUF, the goal of our attacks is to construct a computer algorithm which behaves indistinguishably from the original PUF on almost all CRPs. If successful, this algorithm can subsequently impersonate the Strong PUF, and can be cloned and distributed arbitrarily. It breaks the security of any applications that rest on the Strong PUF’s unpredictability and physical unclonability. Our method is less relevant for other PUF types such as Weak PUFs; see Section I-B for a detailed discussion of this topic. The Strong PUFs that we could attack successfully include standard Arbiter PUFs of essentially arbitrary sizes, and XOR Arbiter PUFs, Lightweight Secure PUFs, and Feed-Forward Arbiter PUFs up to certain sizes and complexities. We also investigate the hardness of certain Ring Oscillator PUF architectures in typical Strong PUF applications. Our attacks are based upon various machine learning techniques, including a specially tailored variant of Logistic Regression and Evolution Strategies. Our results are mostly obtained on CRPs from numerical simulations that use established digital models of the respective PUFs. For a subset of the considered PUFs — namely standard Arbiter PUFs and XOR Arbiter PUFs — we also lead proofs of concept on silicon data from both FPGAs and ASICs. Over four million silicon CRPs are used in this process. The performance on silicon CRPs is very close to simulated CRPs, confirming a conjecture from earlier versions of this work. Our findings lead to new design requirements for secure electrical Strong PUFs, and will be useful to PUF designers and attackers alike.

/pdf/puf-modeling-attacks-on-simulated-and-silicon-data-31z1x12dbl.pdf

PUF Modeling Attacks on Simulated and Silicon Data.

Physically unclonable functions (PUFs) are innovative physical security primitives that produce unclonable and inherent instance-specific measurements of physical objects; in many ways they are the inanimate equivalent of biometrics for human beings. Since they are able to securely generate and store secrets, they allow us to bootstrap the physical implementation of an information security system. In this book the author discusses PUFs in all their facets: the multitude of their physical constructions, the algorithmic and physical properties which describe them, and the techniques required to deploy them in security applications. The author first presents an extensive overview and classification of PUF constructions, with a focus on so-called intrinsic PUFs. He identifies subclasses, implementation properties, and design techniques used to amplify submicroscopic physical distinctions into observable digital response vectors. He lists the useful qualities attributed to PUFs and captures them in descriptive definitions, identifying the truly PUF-defining properties in the process, and he also presents the details of a formal framework for deploying PUFs and similar physical primitives in cryptographic reductions. The author then describes a silicon test platform carrying different intrinsic PUF structures which was used to objectively compare their reliability, uniqueness, and unpredictability based on experimental data. In the final chapters, the author explains techniques for PUF-based entity identification, entity authentication, and secure key generation. He proposes practical schemes that implement these techniques, and derives and calculates measures for assessing different PUF constructions in these applications based on the quality of their response statistics. Finally, he presents a fully functional prototype implementation of a PUF-based cryptographic key generator, demonstrating the full benefit of using PUFs and the efficiency of the processing techniques described. This is a suitable introduction and reference for security researchers and engineers, and graduate students in information security and cryptography.

Physically Unclonable Functions: Constructions, Properties and Applications

This paper introduces a new architecture for circuit-based Physical Unclonable Functions (PUFs) which we call the Bistable Ring PUF (BR-PUF). Based on experimental results obtained from FPGA-based implementations of the BR-PUF, the quality of this new design is discussed in different aspects, including uniqueness and reliability. On the basis of the observed complexity in the challenge-response behavior of BR-PUFs, we argue that this new PUF could be a promising candidate for Strong PUFs. Our design shows noticeable temperature sensitivity, but we discuss how this problem can be addressed by additional hardware and protocol measures.

The Bistable Ring PUF: A new architecture for strong Physical Unclonable Functions

This paper proposes a new approach for the construction of highly secure physical unclonable functions (PUFs). Instead of using systems with medium information content and high readout rates, we suggest to maximize the information content of the PUF while strongly reducing its readout frequency. We show that special, passive crossbar arrays with a very large random information content and inherently limited readout speed are suited to implement our approach. They can conceal sensitive information over long time periods and can be made secure against invasive physical attacks. To support our feasibility study, circuit-level simulations and experimental data are presented. Our design allows the first PUFs that are secure against computationally unrestricted adversaries, and which remain so in the face of weeks or even years of uninterrupted adversarial access. We term the new design principle a “SHIC PUF,” where the acronym SHIC stands for super high information content.

Applications of High-Capacity Crossbar Memories in Cryptography

We observe a security issue in protocols for session key exchange that are based on Strong Physical Unclonable Functions (PUFs). The problem is illustrated by cryptanalyzing a recent scheme of Tuyls and Skoric [1], which has been proposed for use in a bank card scenario. Under realistic assumptions, for example that the adversary Eve can eavesdrop the communication between the players and gains physical access to the PUF twice, she can derive previous session keys in this scheme. The observed problem seems to require the introduction of a new PUF variant, so-called "Erasable PUFs". Having defined this new primitive, we execute some first steps towards its practical implementation, and argue that Erasable PUFs could be implemented securely via ALILE-based crossbar structures.

An attack on PUF-Based session key exchange and a hardware-based countermeasure: erasable PUFs

Diodes are among the most simple and inexpensive electric components. In this paper, we investigate how random diodes with irregular I(U) curves can be employed for crypto and security purposes. We show that such diodes can be used to build Strong Physical Unclonable Functions (PUFs), Certificates of Authenticity (COAs), and Physically Obfuscated Keys (POKs), making them a broadly usable security tool. We detail how such diodes can be produced by an efficient and inexpensive method known as ALILE process. Furthermore, we present measurement data from real systems and discuss prototypical implementations. This includes the generation of helper data as well as efficient signature generation by elliptic curves and 2D barcode generation for the application of the diodes as COAs.

Security applications of diodes with unique current-voltage characteristics

We prepared ultrathin polycrystalline silicon layers (5–100nm) by the aluminum-induced layer exchange process. An Al/oxide/amorphous Si layer stack was annealed at temperatures below 577°C, leading to a layer exchange and the crystallization of the silicon. The process dynamics, structural and electronic properties have been studied. In addition to the well known dependence on the annealing temperature, we found an increase of the nucleation density with layer thickness. Raman spectroscopy shows a good crystalline quality down to a layer thickness of 10nm. Hole concentrations of the p-type layers are between 5×1018 and 9×1019cm−3, depending on layer thickness and annealing temperature.

Structural and electronic properties of ultrathin polycrystalline Si layers on glass prepared by aluminum-induced layer exchange

In this paper, we report on high-rectification pn-diodes (rectification ratios up to 2×107) prepared by aluminum-induced crystallization on crystalline Si-wafers, which exhibit highly random I(V) characteristics. We argue that arrays of such diodes can be employed as physical uncloneable functions for cryptography. To resolve the structure of the active diode area, focused-ion beam imaging was used. The I(V) curves of the diodes reveal that both a smaller polycrystalline silicon film thickness and a smaller diode size lead to increasing randomness due to the increasing inhomogeneity of thinner films and due to more pronounced grain boundary effects for smaller diodes.

C. Jaeger

Papers

Applications of High-Capacity Crossbar Memories in Cryptography

An attack on PUF-Based session key exchange and a hardware-based countermeasure: erasable PUFs

Security applications of diodes with unique current-voltage characteristics

Structural and electronic properties of ultrathin polycrystalline Si layers on glass prepared by aluminum-induced layer exchange

Random pn-junctions for physical cryptography