Testing Physically Unclonable Functions Using Statistical Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for testing the uniqueness and randomness of Physical Unclonable Functions (PUFs) are costly and impractical, as they require querying all instances of a PUF batch, which is not feasible in online and embedded testing scenarios, especially when PUFs are dispersed across different devices and locations.
Innovation Solution
A method and system that allow for the testing of uniqueness and randomness of a PUF instance without needing to access all instances of the batch, by assessing the PUF elements within a single instance, leveraging the inherent correlations between neighboring PUFs due to manufacturing processes, using statistical tests like the Maurer Universal test to determine the uniqueness and randomness indicators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all PUF instances in a batch are queried to test uniqueness and randomness, then accurate assessment of PUF quality is achieved, but the testing becomes costly and impractical
Solution Approach 1:
The patent segments the PUF batch into representative samples and uses statistical testing methods to assess only a subset of PUF instances. By applying the Maurer Universal test and other statistical analyses to a representative sample rather than all instances, the system achieves accurate quality assessment while maintaining high testing efficiency and productivity.
Solution Approach 2:
The patent creates a virtual model of PUF batch characteristics through statistical sampling and mathematical modeling. By analyzing the statistical properties of a representative sample and using these to infer characteristics of the entire batch, the system avoids the need to physically test every PUF instance while maintaining assessment accuracy.
2Reliability
If all PUF instances are accessed for testing, then complete uniqueness verification is achieved, but the complexity and cost increase significantly
Solution Approach 1:
The patent applies partial action by testing only a representative sample of PUF instances rather than all instances in the batch. By carefully selecting samples that represent the diversity of the batch and using statistically significant sample sizes, the system achieves reliable uniqueness verification without the excessive complexity of testing every single instance.
Solution Approach 2:
The patent replaces the mechanical approach of physically accessing and testing every PUF instance with a statistical and mathematical modeling approach. By using statistical tests, sampling theory, and mathematical inference, the system achieves complete verification results without the operational complexity of exhaustive physical testing.
3Ease of operation
If PUFs are tested in online and embedded environments, then practical security testing is enabled, but access to all PUF instances becomes infeasible
Solution Approach 1:
The patent segments the testing approach into sampling and statistical analysis components, enabling online and embedded testing of representative PUF instances without requiring access to all instances. This segmentation allows practical security testing in constrained environments while maintaining statistical validity of the results.
Solution Approach 2:
The patent introduces statistical sampling and mathematical modeling as intermediary mechanisms between the tester and the PUF batch. These intermediaries allow the system to infer characteristics of the entire batch from a representative sample, enabling practical testing in online and embedded environments where accessing all instances is infeasible.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
There is provided a method for testing a Physically Unclonable Function (PUF) implemented in a device, said PUF being configured to receive at least one challenge, each challenge comprising a set of bits, and to produce a set of responses, each response comprising at least one bit and corresponding to one challenge, said PUF comprising a circuitry including a set of PUF elements, each PUF element being controlled by at least one input bit corresponding to at least one bit of said challenge, wherein the method comprises the steps of: - applying at least one bit of the challenge to the PUF instance; - determining (300) identifiers for at least some of the PUF elements, the identifier of each PUF element being determined from the response output by said PUF element in response to said at least one bit of the challenge; - applying a statistical randomness test (304) to a group of identifiers comprising at least some of the identifiers determined for said PUF elements, which provides a test indicator; and - testing said PUF based on said test indicator.