PUF Modeling Attacks on Simulated and Silicon Data
The DOI matches its Crossref record; 47 of the 49 references it cites were found in a registry or an index.
Abstract
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.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“However, this approach has also been shown to be vulnerable to modelling attacks that can be used to infer characteristics well enough to support cloning and redistribution [53].”
From Zero Trust Architecture (ZTA): A Comprehensive Survey (2022) · cites this paper as [53]
Citations
Cites
Book Review: Pattern Recognition and Machine Learning
Nasrabadi, N. M. (2007) · Journal of Electronic Imaging · doi:10.1117/1.2819119 · Cited by 22,066
Pattern Recognition and Machine Learning
(2006) · doi:10.1007/978-0-387-45528-0 · Cited by 9,824
Bäck, T. (1996) · Oxford University Press eBooks · doi:10.1093/oso/9780195099713.001.0001 · Cited by 5,113
A direct adaptive method for faster backpropagation learning: the RPROP algorithm
Riedmiller, M., & Braun, H. (2002) · IEEE International Conference on Neural Networks · doi:10.1109/icnn.1993.298623 · Cited by 3,917
Physical unclonable functions for device authentication and secret key generation
Suh, G. E., & Devadas, S. (2007) · Proceedings of the 44th annual conference on Design automation - DAC '07 · doi:10.1145/1278480.1278484 · Cited by 2,065 · cited 8×
“1) Arbiter PUFs: Arbiter PUFs (Arb-PUFs) were first introduced in [17], [23], [46].”
BackgroundMethodInfluentialSelf-citationPappu, R., Recht, B., Taylor, J., & Gershenfeld, N. (2002) · Science · doi:10.1126/science.1074376 · Cited by 2,018
Learnability and the Vapnik-Chervonenkis dimension
Blumer, A., Ehrenfeucht, A., Haussler, D., & Warmuth, M. K. (1989) · Journal of the ACM · doi:10.1145/76359.76371 · Cited by 1,839 · cited 1×
“Furthermore, the evaluated “logic” security represents an upper limit on a PUF’s ML-resilience, at the least in any applications where perfect error correction or fuzzy extractors are utilized to obtain stable responses, such as PUF-based key exchange [11], [4] and oblivious transfer [36], [4]…”
BackgroundFuzzy Extractors: How to Generate Strong Keys from Biometrics and Other Noisy Data
Dodis, Y., Ostrovsky, R., Reyzin, L., & Smith, A. (2008) · SIAM Journal on Computing · doi:10.1137/060651380 · Cited by 1,717
Silicon physical random functions
Gassend, B., Clarke, D., van Dijk, M., & Devadas, S. (2002) · Proceedings of the 9th ACM conference on Computer and communications security · doi:10.1145/586110.586132 · Cited by 1,660
Self-citationFPGA Intrinsic PUFs and Their Use for IP Protection
Guajardo, J., Kumar, S. S., Schrijen, G.-J., & Tuyls, P. (2007) · Lecture Notes in Computer Science · doi:10.1007/978-3-540-74735-2_5 · Cited by 1,203 · cited 3×
“Unfortunately, it can be difficult to uphold this requirement in practice.”
BackgroundExtracting secret keys from integrated circuits
Lim, D., Lee, J. W., Gassend, B., Suh, G. E., van Dijk, M., & Devadas, S. (2005) · IEEE Transactions on Very Large Scale Integration (VLSI) Systems · doi:10.1109/tvlsi.2005.859470 · Cited by 1,047
Self-citationModeling attacks on physical unclonable functions
Rührmair, U., Sehnke, F., Sölter, J., Dror, G., Devadas, S., & Schmidhuber, J. (2010) · Proceedings of the 17th ACM conference on Computer and communications security · doi:10.1145/1866307.1866335 · Cited by 1,021 · cited 2×
“We used three hardware systems to carry out our experiments: A standalone, consumer INTEL Quadcore Q9300, and a comparable consumer AMD Quadcore, both worth less than 1,000 Euros.”
MethodSelf-citationLee, J. W., Lim, D., Gassend, B., Suh, G. E., van Dijk, M., & Devadas, S. (2004) · 2004 Symposium on VLSI Circuits. Digest of Technical Papers (IEEE Cat. No.04CH37525) · doi:10.1109/vlsic.2004.1346548 · Cited by 842 · cited 3×
“(2) From the different optimization methods which we tested in our ML experiments (standard gradient descent, iterative reweighted least squares, RProp [2], [35]), RProp gradient descent performed best.”
MethodSelf-citationEvolution and Optimum Seeking: The Sixth Generation
Schwefel, H. (1993) · John Wiley & Sons, Inc. eBooks · Cited by 652
Extended abstract: The butterfly PUF protecting IP on every FPGA
Kumar, S. S., Guajardo, J., Maes, R., Schrijen, G.-J., & Tuyls, P. (2008) · 2008 IEEE International Workshop on Hardware-Oriented Security and Trust · doi:10.1109/hst.2008.4559053 · Cited by 547 · cited 1×
“1556-6013 © 2013 IEEE IE EE P ro of W eb V er sio n”
Introduction to Hardware Security and Trust
(2012) · Springer eBooks · doi:10.1007/978-1-4419-8080-9 · Cited by 476
Read-Proof Hardware from Protective Coatings
Tuyls, P., Schrijen, G.-J., Škorić, B., van Geloven, J., Verhaegh, N., & Wolters, R. (2006) · Lecture Notes in Computer Science · doi:10.1007/11894063_29 · Cited by 421 · cited 2×
“Most typical Weak PUFs, such as the SRAM PUF [18], Butterfly PUF [22] or Coating PUF [47], remain unaffected by our attacks.”
BackgroundControlled physical random functions
Gassend, B., Clarke, D., van Dijk, M., & Devadas, S. (2003) · 18th Annual Computer Security Applications Conference, 2002. Proceedings. · doi:10.1109/csac.2002.1176287 · Cited by 380
Self-citationDevadas, S., Suh, E., Paral, S., Sowell, R., Ziola, T., & Khandelwal, V. (2008) · 2008 IEEE International Conference on RFID · doi:10.1109/rfid.2008.4519377 · Cited by 339 · cited 1×
“We also executed a very detailed proof of concept for silicon CRPs for the twomost well-studied and commercially most relevant [9], [10] electrical Strong PUF designs, Arbiter PUFs and XOR Arbiter PUFs.”
MethodSelf-citationIdentification and authentication of integrated circuits
Gassend, B., Lim, D., Clarke, D., van Dijk, M., & Devadas, S. (2004) · Concurrency and Computation: Practice and Experience · doi:10.1002/cpe.805 · Cited by 306
Self-citationCloning Physically Unclonable Functions
Helfmeier, C., Boit, C., Nedospasov, D., & Seifert, J.-P. (2013) · 2013 IEEE International Symposium on Hardware-Oriented Security and Trust (HOST) · doi:10.1109/hst.2013.6581556 · Cited by 283 · cited 1×
“1556-6013 © 2013 IEEE IE EE P ro of W eb V er sio n”
Majzoobi, M., Koushanfar, F., & Potkonjak, M. (2008) · 2008 IEEE/ACM International Conference on Computer-Aided Design · doi:10.1109/iccad.2008.4681648 · Cited by 273 · cited 2×
“(2) From the different optimization methods which we tested in our ML experiments (standard gradient descent, iterative reweighted least squares, RProp [2], [35]), RProp gradient descent performed best.”
MethodGassend, B. (2003) · DSpace@MIT (Massachusetts Institute of Technology) · Cited by 257
Slender PUF Protocol: A Lightweight, Robust, and Secure Authentication by Substring Matching
Majzoobi, M., Rostami, M., Koushanfar, F., Wallach, D. S., & Devadas, S. (2012) · 2012 IEEE Symposium on Security and Privacy Workshops · doi:10.1109/spw.2012.30 · Cited by 227
Self-citationThe Bistable Ring PUF: A new architecture for strong Physical Unclonable Functions
Chen, Q., Csaba, G., Lugli, P., Schlichtmann, U., & Ruhrmair, U. (2011) · 2011 IEEE International Symposium on Hardware-Oriented Security and Trust · doi:10.1109/hst.2011.5955011 · Cited by 227 · cited 1×
“Some first attempts into this direction have already been made in [49], [30], [5], [6], but we did not analyze their viability in detail in this work.”
BackgroundMajzoobi, M., Koushanfar, F., & Potkonjak, M. (2016) · Cited by 211
Techniques for Design and Implementation of Secure Reconfigurable PUFs
Majzoobi, M., Koushanfar, F., & Potkonjak, M. (2009) · ACM Transactions on Reconfigurable Technology and Systems · doi:10.1145/1502781.1502786 · Cited by 204 · cited 1×
“The parameter vector encodes the delays for the subcomponents in the Arb-PUF stages, whereas the feature vector is solely a function of the applied bit challenge [24], [27], [26].”
Testing Techniques for Hardware Security
Majzoobi, M., Koushanfar, F., & Potkonjak, M. (2008) · 2008 IEEE International Test Conference · doi:10.1109/test.2008.4700636 · Cited by 198 · cited 4×
“(2) From the different optimization methods which we tested in our ML experiments (standard gradient descent, iterative reweighted least squares, RProp [2], [35]), RProp gradient descent performed best.”
BackgroundMethodInfluentialOn the Foundations of Physical Unclonable Functions.Preprint
Rührmair, U., Sölter, J., & Sehnke, F. (2009) · IACR Cryptology ePrint Archive · Cited by 190
Self-citationFPGA PUF using programmable delay lines
Majzoobi, M., Koushanfar, F., & Devadas, S. (2010) · 2010 IEEE International Workshop on Information Forensics and Security · doi:10.1109/wifs.2010.5711471 · Cited by 189 · cited 1×
“The parameter vector encodes the delays for the subcomponents in the Arb-PUF stages, whereas the feature vector is solely a function of the applied bit challenge [24], [27], [26].”
Self-citationSecurity Based on Physical Unclonability and Disorder
Rührmair, U., Devadas, S., & Koushanfar, F. (2011) · Springer eBooks · doi:10.1007/978-1-4419-8080-9_4 · Cited by 178
Self-citationMachine learning attacks on 65nm Arbiter PUFs: Accurate modeling poses strict bounds on usability
Hospodar, G., Maes, R., & Verbauwhede, I. (2012) · 2012 IEEE International Workshop on Information Forensics and Security (WIFS) · doi:10.1109/wifs.2012.6412622 · Cited by 173 · cited 1×
“Please note that both eavesdropping on responses as well as physical access to the PUF is part of the established, general attackmodel for PUFs.”
BackgroundFault Injection Modeling Attacks on 65 nm Arbiter and RO Sum PUFs via Environmental Changes
Delvaux, J., & Verbauwhede, I. M. R. (2014) · IEEE Transactions on Circuits and Systems I Regular Papers · doi:10.1109/tcsi.2013.2290845 · Cited by 136
Security, Privacy, and Trust in Modern Data Management
Petković, M., & Jonker, W. (2007) · Springer eBooks · doi:10.1007/978-3-540-69861-6 · Cited by 134
Brucker, P. (2004) · Springer eBooks · doi:10.1007/978-3-540-24804-0_3 · Cited by 115
Strong PUFs: Models, Constructions, and Security Proofs
Rührmair, U., Busch, H., & Katzenbeisser, S. (2010) · Information security and cryptography · doi:10.1007/978-3-642-14452-3_4 · Cited by 111 · cited 1×
“1556-6013 © 2013 IEEE IE EE P ro of W eb V er sio n”
Towards Robust Low Cost Authentication for Pervasive Devices
Ozturk, E., Hammouri, G., & Sunar, B. (2008) · 2008 Sixth Annual IEEE International Conference on Pervasive Computing and Communications (PerCom) · doi:10.1109/percom.2008.54 · Cited by 111
Lightweight and Secure PUF Key Storage Using Limits of Machine Learning
Yu, M.-D., M’Raïhi, D., Sowell, R., & Devadas, S. (2011) · Lecture notes in computer science · doi:10.1007/978-3-642-23951-9_24 · Cited by 94
Self-citationApplications of High-Capacity Crossbar Memories in Cryptography
Rührmair, U., Jaeger, C., Bator, M., Stutzmann, M., Lugli, P., & Csaba, G. (2010) · IEEE Transactions on Nanotechnology · doi:10.1109/tnano.2010.2049367 · Cited by 79 · cited 1×
“Also special PUFs with a very high information content (so-called SHIC PUFs [39], [40], [21]) could be an option, but only in such applications where their slow read-out speed and their comparatively large area consumption are no too strong drawbacks.”
BackgroundTowards Hardware-Intrinsic Security: Foundations and Practice
Sadeghi, A., & Naccache, D. (2010) · Cited by 59
Physical Unclonable Functions in Cryptographic Protocols: Security Proofs and Impossibility Results.Preprint
van Dijk, M., & Rührmair, U. (2012) · IACR Cryptology ePrint Archive · Cited by 50
Unconditionally Secure and Universally Composable Commitments from Physical Assumptions
Damgård, I., & Scafuro, A. (2013) · Lecture notes in computer science · doi:10.1007/978-3-642-42045-0_6 · Cited by 40 · cited 1×
“We use the following definitions throughout the paper: The prediction error is the ratio of incorrect responses of the trained ML algorithm when evaluated on the test set.”
MethodSecurity Applications of Diodes with Unique Current-Voltage Characteristics
Rührmair, U., Jaeger, C., Hilgers, C., Algasinger, M., Csaba, G., & Stutzmann, M. (2010) · Lecture notes in computer science · doi:10.1007/978-3-642-14577-3_26 · Cited by 34 · cited 1×
“Also special PUFs with a very high information content (so-called SHIC PUFs [34], [35], [15]) could be an option, but only in such applications where their slow read-out speed and their comparatively large area consumption are no too strong drawbacks.”
BackgroundApplication of mismatched Cellular Nonlinear Networks for Physical Cryptography
Csaba, G., Ju, X., Ma, Z., Chen, Q., Wolfgang Porod, Y., Schmidhuber, J., Schlichtmann, U., Lugli, P., & Rührmair, U. (2010) · International Workshop on Cellular Nanoscale Networks and their Applications (CNNA) · doi:10.1109/cnna.2010.5430303 · Cited by 27
Self-citationRandom pn-junctions for physical cryptography
Jaeger, C., Algasinger, M., Rührmair, U., Csaba, G., & Stutzmann, M. (2010) · Applied Physics Letters · doi:10.1063/1.3396186 · Cited by 25
Characterization of the bistable ring PUF
Qingqing, C., Csaba, G., Lugli, P., Schlichtmann, U., & Rührmair, U. (2012) · Design, Automation, and Test in Europe · doi:10.1109/date.2012.6176596 · Cited by 25
Physical Unclonable Functions and Secure Processors
Devadas, S. (2009) · Lecture notes in computer science · doi:10.1007/978-3-642-04138-9_5 · Cited by 17 · cited 1×
“If cores are used in parallel, the computation times can be reduced roughly by a factor of , since our ML algorithms parallelize straightforwardly.”
MethodSelf-citation
Research works printed but not linked: 8
- pappu, 2001, Physical one-way functions
- r�hrmair, 2010, Oblivious transfer based on physical unclonable functions (extended abstract), Proc TRUST 2010, 6101, 430
- holcomb, 2007, Initial sram state as a fingerprint and source of true random numbers for RFID tags, Proc Conf RFID Security
- lim, 2004, Extracting secret keys from integrated circuits
- 2004, System and Method of Reliable Forward Secret Key Sharing With Physical Random Functions
- brzuska, 0, Physical unclonable functions in the universal composition framework, Proc CRYPTO 2011, 51
- s�lter, 2009, Cryptanalysis of Electrical PUFs via Machine Learning Algorithms
- schaul, 2010, PyBrain, J Mach Learning Res, 1, 999
Cited by
Gao, Y., Al-Sarawi, S. F., & Abbott, D. (2020) · Nature Electronics · doi:10.1038/s41928-020-0372-5 · Cited by 613
Zero Trust Architecture (ZTA): A Comprehensive Survey
Syed, N. F., Shah, S. W., Shaghaghi, A., Anwar, A., Baig, Z., & Doss, R. (2022) · IEEE Access · doi:10.1109/access.2022.3174679 · Cited by 406
“However, this approach has also been shown to be vulnerable to modelling attacks that can be used to infer characteristics well enough to support cloning and redistribution [53].”
BackgroundMcGrath, T., Bagci, I. E., Wang, Z. M., Roedig, U., & Young, R. J. (2019) · Applied Physics Reviews · doi:10.1063/1.5079407 · Cited by 353
“12 Alongside the disagreement as to what constitutes a “weak” and “strong” PUF, there is a level of disagreement as to what can be called a PUF in the first place.”
BackgroundThe Gap Between Promise and Reality: On the Insecurity of XOR Arbiter PUFs
Becker, G. T. (2015) · Lecture Notes in Computer Science · doi:10.1007/978-3-662-48324-4_27 · Cited by 254
“Their initial analysis was conducted on simulated data but they later verified the results by using measurements taken from an ASIC [26].”
MethodA Multiplexer-Based Arbiter PUF Composition with Enhanced Reliability and Security
Sahoo, D. P., Mukhopadhyay, D., Chakraborty, R. S., & Nguyen, P. H. (2018) · IEEE Transactions on Computers · doi:10.1109/tc.2017.2749226 · Cited by 213
“The CM attacks of x-XOR APUF using logistic regression (LR) are reported in [5], [9], [10], and SCM attacks are published in [11]–[15].”
Method
Metadata
- Authors
- Ruhrmair, U., Solter, J., Sehnke, F., Xu, X., Mahmoud, A., Stoyanova, V., Dror, G., Schmidhuber, J., Burleson, W., & Devadas, S.
- Affiliations
- Technical University of Munich (DE)
- University of Massachusetts Amherst (US)
- Academic College of Tel Aviv-Yafo (IL)
- MIT Computer Science and Artificial Intelligence Laboratory (US)
- Venue
- IEEE Transactions on Information Forensics and Security
- Published
- 2013-11
- Volume
- 8
- Issue
- 11
- Pages
- 1876-1891
- ISSN
- 1556-6013 (ISSN-L)
- 1556-6021
- OpenAlex
- W2044389004
- Cited by
- 637
- References
- 49
- Citations by year
- 2026: 39 · 2025: 50 · 2024: 43 · 2023: 62 · 2022: 47 · 2021: 75 · 2020: 63 · 2019: 66 · 2018: 57 · 2017: 49 · 2016: 37 · 2015: 32 · 2014: 12 · 2013: 2
- FWCI
- 51.65
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- IEEE Transactions on Information Forensics and Security (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Hardware and Architecture › Physical Unclonable Functions (PUFs) and Hardware Security
- Physical Sciences › Computer Science › Artificial Intelligence › Cryptographic Implementations and Security
- Physical Sciences › Mathematics › Discrete Mathematics and Combinatorics › Coding theory and cryptography
- Keywords
- physical unclonable functions, arbiter PUF, ring oscillator PUF, challenge-response pairs, logistic regression, evolution strategies, FPGA, ASIC
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.