Homomorphic Encryption for Secure Pattern Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The application of homomorphic encryption in pattern matching is impractical due to high data amounts and calculation loads, making it inefficient for secure pattern matching in cloud environments.
Innovation Solution
The method involves generating polynomials using binary vectors in ascending or descending order and encrypting them using a homomorphic encryption method, such as Ring-LWE, to perform polynomial processing, reducing data size and calculation costs by calculating hamming distances through cryptographic operations without decrypting the data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If homomorphic encryption is applied to pattern matching, then data security is improved, but data size and calculation load increase making it impractical
Solution Approach 1:
The patent segments the pattern matching process into multiple steps: generating binary vectors from data, converting them to polynomials, encrypting the polynomials, performing distance calculations on encrypted data, and finally decrypting results. This segmentation allows applying homomorphic encryption only where necessary while managing data size through structured transformation.
Solution Approach 2:
The patent transforms data from its original form through parameter changes: converting data to binary vectors, then to polynomials with specific degree constraints, and finally to encrypted form. These parameter transformations optimize the data representation for secure processing while controlling the growth of data size through mathematical constraints on polynomial degrees.
2Reliability
If homomorphic encryption is applied to pattern matching, then data security is improved, but calculation load increases making it impractical
Solution Approach 1:
The patent extracts only the essential computational operations needed for pattern matching and implements them in encrypted space. By taking out the distance calculation step and performing it directly on encrypted polynomials using homomorphic properties, the system avoids decrypting data for computation, thus improving security while managing calculation load through targeted cryptographic operations.
Solution Approach 2:
The patent performs preliminary actions by pre-generating binary vectors and converting them to polynomials before encryption. This preliminary structuring of data in polynomial form allows efficient homomorphic operations during the encrypted distance calculation phase, reducing the overall calculation load compared to performing conversions after encryption.
3Productivity
If traditional pattern matching is used, then calculation speed is maintained, but data security is compromised due to plaintext processing
Solution Approach 1:
The patent introduces encrypted polynomials as an intermediary representation between plaintext data and the pattern matching operation. This intermediary form allows distance calculations to be performed without exposing the actual data values, maintaining security while enabling computation through homomorphic properties that preserve the ability to calculate Hamming distances.
Data Source
Figure 1~3
Figure 2(A)~2(D)
Figure 4
AI summary
An encrypted first polynomial that is obtained by encrypting, in a homomorphic encryption method that handles a polynomial processing, a first polynomial, is received from another computer. The first polynomial is represented by using, as coefficients, components of a first binary vector generated from first data in first order that is either ascending order or descending order with respect to degree of the first polynomial. Then, a predetermined processing in an encrypted text space is performed by using the encrypted first polynomial and an encrypted second polynomial that is obtained by encrypting a second polynomial in the homomorphic encryption method. The second polynomial is represented by using, as coefficients, components of a second binary vector generated from second data in second order that is different from the first order with respect to degree of the second polynomial. Then, a result of the predetermined processing is sent back.