Computation Device Using Shared Shares for Secure Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing white-box implementations are vulnerable to share reduction and share localization attacks, which compromise the security of computations involving sensitive data by reducing the entropy introduced by share representations.
Innovation Solution
A computation device and method that perform operations on shares of values, where each value is stored as a plurality of shares, with a shared share and a computed share being used to define an output value, making it difficult for attackers to determine the shares and introducing faults that affect the computation differently, thereby complicating attacks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If shares are used to represent sensitive values, then protection against side channel attacks is improved, but vulnerability to share reduction and share localization attacks increases
Solution Approach 1:
The patent divides sensitive values into multiple shares, where each share alone reveals no information about the original value. This segmentation approach protects against side channel attacks by ensuring that even if one share is compromised, the complete sensitive information remains protected. The segmentation is applied throughout the computation process, with intermediate values also being shared.
Solution Approach 2:
The patent introduces dummy shares as intermediaries that are computationally mixed with real shares during operations. These dummy shares act as mediators that obscure the relationship between input shares and output shares, making it difficult for attackers to perform share reduction or share localization attacks while preserving the security benefits of share representation.
2Reliability
If multiple shares are used to define each value, then entropy and security are increased, but complexity of determining shares and computing operations increases
Solution Approach 1:
The computation process is segmented into operations that work independently on shares without requiring reconstruction of the original sensitive values. This allows security to be maintained throughout the computation while avoiding the complexity of share reconstruction and management.
Solution Approach 2:
The patent designs computation operations that can handle multiple types of values (sensitive and non-sensitive) using the same share-based approach. This universal method simplifies the overall system complexity by providing a unified framework for secure computation across different data types and operations.
3Object-affected harmful factors
If shared shares are used that also define further values, then fault introduction becomes more difficult to detect, but computation complexity increases
Solution Approach 1:
The patent merges the representation of multiple values into shared shares, where a single share can contribute to multiple output values. This merging approach increases the difficulty of detecting faults because a fault in a shared share can propagate to multiple values in unpredictable ways, while the computational overhead is managed through efficient share manipulation operations.
Data Source
AI summary
Some embodiments are directed to a computation device for performing a computation on at least a set of values. The values are stored in memory as a plurality of shares that define the value. An operation of the computation may be performed on a set of input values to obtain an output value. The output value may be defined by at least one shared share and at least one computed share. The at least one shared share may also define a further value, e.g., an output of a previously performed computation or an output of a further operation performed in parallel with the operation. The at least one computed share is computed from the at least one shared share and shares of the set of input values. A fault in the shared share affects the further value but a fault in the computed share, complicating share reduction attacks.