Polynomial Coefficient Encoding for Secure Distributed Computation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distributed computing systems face high bandwidth demands when performing secure computations, particularly in operations like secure equality protocols and computations involving products or functions of data from multiple databases, leading to inefficient communication and increased bandwidth usage.
Innovation Solution
An asymmetric operation method where devices encode and transmit encrypted coefficients of specific polynomials to perform secure computations, reducing the need for direct data transmission by using secure multiparty equality and multiplication protocols, and employing selection filters to minimize communication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If secure equality protocols are used to compare data from multiple databases, then data security is improved, but communication bandwidth increases significantly
Solution Approach 1:
The patent segments the data comparison process into two phases: a filtering phase where databases are pre-processed to identify candidate records, and a verification phase where only candidates undergo secure equality protocols. This segmentation reduces the number of records requiring full secure comparison from N×M to a smaller subset, thereby reducing communication bandwidth while maintaining security.
Solution Approach 2:
The patent applies preliminary actions by executing filtering operations before the secure equality protocols. Databases are pre-processed to eliminate obviously non-matching records based on indexing or preliminary comparisons, so that only a reduced set of candidate records proceeds to the bandwidth-intensive secure verification phase. This preliminary action significantly reduces the effective N×M communication burden.
2Productivity
If direct data transmission is used for computations, then computation efficiency is improved, but data security deteriorates
Solution Approach 1:
The patent introduces encrypted data and intermediate results as intermediaries between the databases and the computation results. Data is encrypted before transmission and processing, and only decrypted at the final output stage. This intermediary approach allows computations to proceed efficiently while maintaining security, as the intermediary encrypted forms enable processing without exposing plain data throughout the computation pipeline.
3Measurement precision
If all data records are compared in secure equality protocols, then computation completeness is improved, but communication complexity increases proportionally to N×M
Solution Approach 1:
The patent applies local quality by treating different records differently based on their characteristics. Records are localized into candidate sets based on their similarity or matching potential, and only these localized candidates undergo full secure comparison. This allows the system to maintain computation completeness for relevant data while avoiding the N×M complexity of comparing all records pairwise, as unrelated records are excluded from the verification phase.
Data Source
AI summary
Computation efficiency of distributed secure implementation of the computation of a (sum of) products of values Vi, Wi from different servers on a distributed computing system is improved by generation of coefficients of a first and second polynomials P, Q by a first server. The first polynomial P has all numbers Xi from a first data set on the first server as roots. The second polynomial Q has values Q(Xi)=Vi for the numbers Xi from the first data set. The first server transmits coefficients of the polynomials to a second server in encrypted form. The second sever computes encrypted values <P(Xi′)> and <Q(Xi′)> of the polynomials for a number Xi′ in a second set from the encrypted coefficients. The second server computes an encrypted binary value <di> from the encrypted value <p(Xi′) of the first polynomial p and computes an encrypted value of a product <di Q(Xi′) Wi>.