Biometric Database Query Response Randomization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric databases are vulnerable to query attacks where attackers can determine if an individual is a passenger by repeatedly querying biometric data points, compromising security.
Innovation Solution
Implement a biometric database system where reference biometric data is associated with disjoint sets of random variables, generating responses containing random variables that satisfy specific conditions, making it difficult for attackers to infer database contents.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a biometric database stores reference biometric data points for identity verification, then identity verification capability is improved, but the database becomes vulnerable to query attacks where attackers can determine if an individual is referenced by repeatedly querying biometric data points
Solution Approach 1:
The database associates each reference biometric data point with a disjoint set of random variables. When a query is received, the system segments the response generation by selecting random variables from the appropriate set based on whether the biometric data point is found, rather than returning a simple yes/no answer. This segmentation of the response space prevents attackers from learning database contents through repeated queries.
Solution Approach 2:
Random variables serve as an intermediary between the actual biometric database contents and the query responses. Instead of directly revealing whether a biometric data point exists in the database, the system uses random variables as a mediator to provide verification information without exposing sensitive database contents to potential attackers.
2Speed
If the database returns straightforward yes/no responses to queries, then query processing speed is improved, but information leakage occurs that allows attackers to infer database contents
Solution Approach 1:
The system changes the parameter space of the response from simple binary (yes/no) to a larger space of possible random variable values. By transforming the response format while maintaining the core verification functionality, the system prevents information leakage without significantly impacting query processing speed, as the random variable selection and comparison operations remain computationally efficient.
Data Source
Figure 1
Figure 2
AI summary
A method comprising: receiving (208) a first biometric data point and a query asking whether the first biometric data point is referenced in a database (DBB), the database comprising reference biometric data points associated with respective sets of random numbers that are disjoint; searching (210) for a second biometric data point corresponding to the first biometric data point in the database; generating (212) a response to the query containing a random number satisfying the following conditions: when the second biometric data point is found during the search, the random number is an element of the set of random numbers associated with the second biometric data point; when the second biometric data point is not found during the search, the random number does not belong to any of the sets of random numbers; the random number is different from a previous random number contained in a previous response to a previous query asking whether a previous biometric data point is referenced in the database.