Hidden Markov Model Authentication for Correlated PUF Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing authentication methods using Physically Unclonable Functions (PUFs) face challenges in accurately identifying devices with temporally correlated data, as traditional statistical processing is inefficient and prone to errors due to interdependence of data bits, particularly in small data volumes like those generated by random number PUFs.
Innovation Solution
The implementation of a hidden Markov model (HMM) for authentication, combined with error correction techniques like the Bose-Chanduri-Hocquenghem (BCH) or Reed-Solomon codes, to generate and verify device-specific IDs, effectively handling correlated data and correcting errors in small data sets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional statistical processing is used for authentication, then the method is simple to implement, but authentication accuracy deteriorates due to inefficiency and errors from interdependence of data bits
Solution Approach 1:
The patent introduces a Markov chain model as an intermediary between the correlated data bits and the authentication decision. This model captures the temporal dependencies and statistical characteristics of the data without requiring complex direct analysis, thereby improving authentication accuracy while maintaining manageable processing complexity through probabilistic modeling.
Solution Approach 2:
The patent transforms the authentication approach by changing from traditional statistical processing parameters to Markov chain parameters (transition probabilities, state definitions). This parameter transformation enables the system to handle temporally correlated data effectively, improving measurement precision while the computational overhead remains acceptable for practical authentication scenarios.
2Reliability
If data bits are processed independently, then processing is simplified, but authentication reliability deteriorates due to interdependence of data bits in small data volumes
Solution Approach 1:
The patent applies dynamics by modeling the temporal evolution of data bits through a Markov chain, where the state at any time step depends on previous states. This dynamic model captures the interdependence of data bits naturally, improving authentication reliability for small data volumes without requiring complex static analysis of all bit combinations.
Solution Approach 2:
The Markov chain model incorporates feedback mechanisms where the probability distribution at each time step is updated based on observed data sequences. This feedback allows the system to adapt to the specific statistical characteristics of the data source, improving reliability while the computational complexity is managed through efficient probability calculations rather than exhaustive analysis.
3Measurement precision
If small data volumes are used, then device complexity is reduced, but authentication accuracy deteriorates due to insufficient data for statistical processing
Solution Approach 1:
The patent applies preliminary action by pre-defining the Markov chain state space and transition probability models before authentication occurs. This allows the system to extract meaningful authentication information from small data volumes by leveraging the pre-established probabilistic framework, which guides the analysis and reduces the amount of data needed for accurate authentication while maintaining low device complexity.
Data Source
AI summary
According to an embodiment, an authentication system includes a physical device, a calculator, and an authenticator. The physical device includes a data source which outputs a data sequence along time series. The calculator performs, using hidden Markov model, probability calculation on an ID which is based on the data sequence obtained from the physical device. The authenticator authenticates the physical device based on calculation result of the calculator.