Secure Permutation Protocol for Encrypted Data Intersection Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current secure multi-party computing (MPC) protocols face challenges in efficiently matching intersections between dispersed data sources while ensuring data security and compliance with regulations, particularly due to high computational and communication costs, especially when dealing with unbalanced datasets and duplicate IDs.
Innovation Solution
The proposed method employs a secure permutation protocol to generate feature shares of intersection data without exposing the intersection, using double-encrypted identification information and feature shares to align datasets, allowing for efficient MPC operations without revealing sensitive information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional MPC protocols are used for intersection matching between dispersed data sources, then data security is maintained, but computational cost and communication cost increase significantly
Solution Approach 1:
The patent segments the intersection matching process into multiple rounds of secure permutation operations. Instead of performing a single complex MPC intersection protocol, the system divides the computation into iterative steps where feature shares are progressively aligned and matched, reducing the computational burden in each individual round while maintaining overall security
Solution Approach 2:
The patent performs preliminary actions by pre-processing data entries into feature shares before the actual intersection matching. The double-encrypted identification information is prepared in advance, and feature shares are generated and distributed beforehand, allowing the MPC protocol to operate on pre-processed data structures that reduce computational complexity during the matching phase
2Reliability
If traditional MPC protocols are used for intersection matching, then data security is ensured, but communication cost increases
Solution Approach 1:
The patent extracts and processes only the necessary feature shares for intersection matching rather than transmitting entire datasets. By working with extracted feature shares and double-encrypted identification information, the system reduces communication volume to only the essential data elements needed for secure intersection computation
Solution Approach 2:
The patent transforms the intersection matching problem from operating on raw data entries to operating on feature shares in a different dimensional space. This dimensionality change allows multiple data entries to be represented and processed through their feature share representations, reducing the communication overhead while preserving the ability to identify intersections
3Measurement precision
If datasets are aligned for MPC operations, then intersection matching accuracy improves, but the amount of data to be buffered increases
Solution Approach 1:
The patent applies local quality by aligning only the feature shares relevant to intersection matching rather than buffering entire datasets. The system maintains local alignments of identification information and corresponding feature shares, processing data in localized segments that require minimal buffering while achieving accurate intersection identification
Data Source
AI summary
Data processing methods and electronic device are provided in the present disclosure. A method comprises: obtaining, by a first party, first double-encrypted identification information and a first feature share, a second feature share of first feature information in a first dataset and second double-encrypted identification information and a third feature share of second feature information in a second dataset; determining a first intersection of the first and second datasets by matching the first double-encrypted identification information with the second double-encrypted identification information; performing, using the first feature share, a first round of secure permutation with the second party, to obtain a first target feature share of the first feature information; performing, using the third feature share, a second round of secure permutation with the second party, to obtain a first target feature share of the second feature information; updating the first intersection and performing MPC with the second party.