Encrypted Similarity Calculation via Homomorphic Encryption

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Authentication systems using additive homomorphic encryption face challenges in preventing information leakage and spoofing attacks, particularly in maintaining the secrecy of original data during operations between apparatuses.

Innovation Solution

A similarity degree calculation apparatus that processes and compares encrypted data using a temporary key, ensuring that the similarity degree calculation remains encrypted throughout the process, thereby preventing information leakage and enabling secure authentication.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If additive homomorphic encryption is used to perform arithmetic operations on encrypted data, then the ability to calculate similarity degrees on encrypted data is improved, but the risk of information leakage and spoofing attacks increases

Engineering Contradiction:
Improveability to calculate similarity on encrypted dataVSAvoidinformation leakage and spoofing attacks
Core Design Contradiction:
Extent of automationVSObject-affected harmful factors

Solution Approach 1:

The patent introduces a trusted third party (server) that holds the secret key and performs decryption operations. This intermediary enables the system to benefit from homomorphic encryption's computational capabilities while mitigating security risks by centralizing secret key management and preventing direct exposure of encrypted data between client devices

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system divides the encryption key into two parts: a public key held by client devices for encryption, and a secret key held by the server for decryption. This segmentation allows encrypted data to be processed and compared while maintaining security through separated key management responsibilities

Inventive Principle:
Principle #1Segmentation

2Reliability

If encrypted data is transmitted and processed between apparatuses, then data secrecy is improved, but the complexity of the authentication system increases

Engineering Contradiction:
Improvedata secrecyVSAvoidauthentication system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

Client devices perform self-encryption of data using the public key before transmission, and the server performs self-decryption using the secret key. This self-service approach to encryption/decryption simplifies the overall system architecture by eliminating the need for complex key exchange protocols and mutual authentication mechanisms

Inventive Principle:
Principle #25Self-service

3Reliability

If homomorphic encryption is used to protect original data secrecy, then security against spoofing attacks is improved, but the operational efficiency and speed of similarity calculation deteriorates

Engineering Contradiction:
Improvesecurity against spoofing attacksVSAvoidsimilarity calculation speed
Core Design Contradiction:
ReliabilityVSSpeed

Solution Approach 1:

The system performs partial decryption by the server (decrypting only the encrypted data received from clients, not all data in the system) and then transfers the decrypted data to client devices for final similarity calculation. This partial action approach balances security requirements with computational efficiency by limiting the server's processing burden while maintaining data protection

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP2680488B1Similarity calculation system, similarity calculation device, computer program, and similarity calculation method
Publication Date: 2019.08.21 MITSUBISHI ELECTRIC CORP
  • EP2680488B1 patent drawingFigure 1~2
  • EP2680488B1 patent drawingFigure 3~4
  • EP2680488B1 patent drawingFigure 5

AI summary

Based on an encrypted feature vector C (comparison ciphertext) encrypted with a public key pk of a decryption apparatus and an encrypted feature vector C' (target ciphertext) encrypted with the public key pk of the decryption apparatus, and a random number (temporary key) generated by a random number generation unit 303 (temporary key generation unit), an encrypted random similarity degree calculation unit 314 (interim similarity degree ciphertext calculation unit) performs calculation for calculationg a similarity degree in a first stage, with two encrypted feature vectors C and C' kept encrypted, thereby calculating a second challenge C". The decryption apparatus decrypts the second challenge C^ with a secret key sk of the decryption apparatus, and performs calculation for calculating the similarity degree in a second stage with a result of the decryption kept encrypted with the temporary key, thereby calculating a second response Z. A plaintext similarity degree extraction unit 315 (similarity degree calculation unit) decrypts the second response Z with the temporary key, thereby calculating a similarity degree d.