Design Automation and Test in Europe 2014 PUFs at a Ulrich Rührmair � Technische Universität München � � Glance Daniel E. Holcomb � University of Michigan This work was supported in part by C-FAR, one of six centers of STARnet, a Semiconductor Research Corporation program sponsored by MARCO and DARPA, and NSF CNS-0845874.
Physical Unclonable Functions Silicon Physical Random Functions ∗ Blaise Gassend, Dwaine Clarke, Marten van Dijk † and Srinivas Devadas Massachusetts Institute of Technology Laboratory for Computer Science Cambridge, MA 02139, USA gassend,declarke,marten,devadas @mit.edu ABSTRACT Research � We introduce the notion of a Physical Random Function Mentions � (PUF). We argue that a complex integrated circuit can be viewed as a silicon PUF and describe a technique to identify by Year and authenticate individual integrated circuits (ICs). We describe several possible circuit realizations of di ff er- Year � 2 PUFs at a Glance DATE 2014
Overview Context and motivation for remainder of session 1. Brief introduction to PUFs � 2. Weak PUFs and applications � 3. Strong PUFs and applications � 4. Conclusions � 3 PUFs at a Glance DATE 2014
Physical Unclonable Functions Challenges Responses f � 4 PUFs at a Glance DATE 2014
Physical Unclonable Functions Challenges Responses ❖ Function � f ❖ Map challenges to responses � 4 PUFs at a Glance DATE 2014
Physical Unclonable Functions Challenges Responses ❖ Function � f ❖ Map challenges to responses ❖ Physical � ❖ Mapping depends on physical variations � 4 PUFs at a Glance DATE 2014
Physical Unclonable Functions Challenges Responses ❖ Function � f ❖ Map challenges to responses ❖ Physical � ❖ Mapping depends on physical variations � 4 PUFs at a Glance DATE 2014
Physical Unclonable Functions Challenges Responses ❖ Function � f ❖ Map challenges to responses PUF Characterized by Challenge-Response Pairs ❖ Physical � (CRPs) ❖ Mapping depends on physical variations � 4 PUFs at a Glance DATE 2014
Physical Unclonable Functions Challenges Responses ❖ Function � f ❖ Map challenges to responses PUF Characterized by Challenge-Response Pairs ❖ Physical � (CRPs) ❖ Mapping depends on physical variations ❖ Unclonable � ❖ No compact model exists, and CRP space is too large for dictionary � 4 PUFs at a Glance DATE 2014
Physical Unclonable Functions Challenges Responses ❖ Function � f ❖ Map challenges to responses PUF Characterized by Challenge-Response Pairs ❖ Physical � (CRPs) ❖ Mapping depends on physical variations ❖ Unclonable � ❖ No compact model exists, and CRP space is too large for dictionary ❖ Or, responses kept secret � 4 PUFs at a Glance DATE 2014
Design Considerations for Silicon PUFs ❖ Outputs determined by uncorrelated variation � ❖ Random dopant fluctuations and small devices � ❖ Balanced parasitics and wire lengths to avoid bias � 5 PUFs at a Glance DATE 2014
Design Considerations for Silicon PUFs ❖ Outputs determined by uncorrelated variation � ❖ Random dopant fluctuations and small devices � ❖ Balanced parasitics and wire lengths to avoid bias ❖ Variation and noise hard to separate � ❖ Mask unreliable outputs � ❖ Majority voting � ❖ Error correction � 5 PUFs at a Glance DATE 2014
Design Considerations for Silicon PUFs ❖ Outputs determined by uncorrelated variation � ❖ Random dopant fluctuations and small devices � ❖ Balanced parasitics and wire lengths to avoid bias ❖ Variation and noise hard to separate � ❖ Mask unreliable outputs � ❖ Majority voting � ❖ Error correction ❖ Secure � 5 PUFs at a Glance DATE 2014
Security Considerations ❖ Assumed capabilities of adversary � ❖ Observe CRPs � ❖ Measure side channels � ❖ Disassemble and probe chip � 6 PUFs at a Glance DATE 2014
Security Considerations ❖ Assumed capabilities of adversary � ❖ Observe CRPs � ❖ Measure side channels � ❖ Disassemble and probe chip ❖ Possible results of attacks � ❖ DOS by increasing error rate of CRPs � ❖ Train parametric model to predict responses � ❖ Clone with another instance of PUF � 6 PUFs at a Glance DATE 2014
Security Considerations ❖ Assumed capabilities of adversary � ❖ Observe CRPs � 2 nd talk of session ❖ Measure side channels � ❖ Disassemble and probe chip ❖ Possible results of attacks � ❖ DOS by increasing error rate of CRPs � ❖ Train parametric model to predict responses � ❖ Clone with another instance of PUF � 6 PUFs at a Glance DATE 2014
Security Considerations ❖ Assumed capabilities of adversary � ❖ Observe CRPs � ❖ Measure side channels � 3 rd talk of session ❖ Disassemble and probe chip ❖ Possible results of attacks � ❖ DOS by increasing error rate of CRPs � ❖ Train parametric model to predict responses � ❖ Clone with another instance of PUF � 6 PUFs at a Glance DATE 2014
Security Considerations ❖ Assumed capabilities of adversary � ❖ Observe CRPs � ❖ Measure side channels � ❖ Disassemble and probe chip 4 th talk of session ❖ Possible results of attacks � ❖ DOS by increasing error rate of CRPs � ❖ Train parametric model to predict responses � ❖ Clone with another instance of PUF � 6 PUFs at a Glance DATE 2014
Weak vs Strong PUFs Weak PUFs � Strong PUFs � � 7 PUFs at a Glance DATE 2014
Weak vs Strong PUFs Weak PUFs � Strong PUFs � ❖ Few/one challenges ❖ Many challenges � 7 PUFs at a Glance DATE 2014
Weak vs Strong PUFs Weak PUFs � Strong PUFs � ❖ Few/one challenges ❖ Many challenges ❖ Responses remain internal ❖ Public CRP interface ❖ Perfect internal error ❖ Error correction outside correction PUF is possible � 7 PUFs at a Glance DATE 2014
Weak vs Strong PUFs Weak PUFs � Strong PUFs � ❖ Few/one challenges ❖ Many challenges ❖ Responses remain internal ❖ Public CRP interface ❖ Perfect internal error ❖ Error correction outside correction PUF is possible ❖ Attacks: Cloning and invasive ❖ Attacks: Modeling attacks reading of responses and protocol attacks � 7 PUFs at a Glance DATE 2014
Weak vs Strong PUFs Weak PUFs � Strong PUFs � ❖ Few/one challenges ❖ Many challenges ❖ Responses remain internal ❖ Public CRP interface ❖ Perfect internal error ❖ Error correction outside correction PUF is possible ❖ Attacks: Cloning and invasive ❖ Attacks: Modeling attacks reading of responses and protocol attacks ❖ Use cases: New form of key storage � 7 PUFs at a Glance DATE 2014
Weak vs Strong PUFs Weak PUFs � Strong PUFs � ❖ Few/one challenges ❖ Many challenges ❖ Responses remain internal ❖ Public CRP interface ❖ Perfect internal error ❖ Error correction outside correction PUF is possible ❖ Attacks: Cloning and invasive ❖ Attacks: Modeling attacks reading of responses and protocol attacks ❖ Use cases: New form of key ❖ Use cases: New cryptographic storage primitive � 7 PUFs at a Glance DATE 2014
Weak vs Strong PUFs Weak PUFs � Strong PUFs � ❖ Few/one challenges ❖ Many challenges ❖ Responses remain internal ❖ Public CRP interface ❖ Weak and strong are two PUF subclasses among many � ❖ Perfect internal error ❖ Error correction outside ❖ Controlled PUFs � correction PUF is possible ❖ Attacks: Cloning and invasive ❖ Public PUFs � ❖ Attacks: Modeling attacks reading of responses and protocol attacks ❖ SIMPL, etc ❖ Use cases: New form of key ❖ Use cases: New cryptographic storage primitive � 7 PUFs at a Glance DATE 2014
Overview 1. Brief introduction to PUFs � 2. Weak PUFs and applications � 3. Strong PUFs and applications � 4. Conclusions � 8 PUFs at a Glance DATE 2014
Examples of Weak PUFs ❖ Using custom circuits � ❖ Drain currents [Lofstrom et al. ’02] � ❖ Capacitive coating PUF [Tuyls et al. ’06] � ❖ Cross-coupled devices [Su et al. ’07] � ❖ Sense amps [Bhargava et al. ’10] � ❖ Using existing circuits � ❖ Clock skew [Yao et al.’13] � ❖ Flash latency [Prabhu et al. ‘11] � ❖ Power-up SRAM state [Guajardo et al. ’07, Holcomb et al. ’07] � 9 PUFs at a Glance DATE 2014
Examples of Weak PUFs ❖ Using custom circuits � ❖ Drain currents [Lofstrom et al. ’02] � ❖ Capacitive coating PUF [Tuyls et al. ’06] � ❖ Cross-coupled devices [Su et al. ’07] � "SRAM PUF" "PUF" ❖ Sense amps [Bhargava et al. ’10] � Research � ❖ Using existing circuits � Mentions � by Year ❖ Clock skew [Yao et al.’13] � ❖ Flash latency [Prabhu et al. ‘11] � Year ❖ Power-up SRAM state [Guajardo et al. ’07, Holcomb et al. ’07] � 9 PUFs at a Glance DATE 2014
Applications of Weak PUFs ❖ Identification � ❖ Authentication � ❖ Secret key � ❖ Random number generation � 10 PUFs at a Glance DATE 2014
Applications of Weak PUFs ❖ Identification � ❖ Authentication � ❖ Secret key � ❖ Random number generation � 10 PUFs at a Glance DATE 2014
SRAM Power-up State Utilize inherent power-up bias of each SRAM cell 1.2 WL VDD A VDD 0.8 B Voltage 0.4 B A BL BLB 0 0 2 4 6 8 10 Time [ns] � 11 PUFs at a Glance DATE 2014
Recommend
More recommend
