Secure Hamming Distance Computation via Oblivious Transfer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing secure methods for calculating Hamming distances in biometric identification, such as garbled circuits and homomorphic encryption, are costly in terms of calculation time and computer resources, and are not optimized for determining multiple Hamming distances in parallel.
Innovation Solution
A secure method involving a client and a server that evaluates a linear combination of sub-functions with binary inputs, using oblivious transfer to compute Hamming distances without either party obtaining information about the other's code, by generating random values and combining them using doublets, allowing for efficient and parallel computation of Hamming distances.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If secure methods such as garbled circuits or homomorphic encryption are used to calculate Hamming distances, then data confidentiality is preserved, but computational cost and time increase significantly
Solution Approach 1:
The method segments the secure computation into distinct phases: the server generates random values and computes intermediate results independently, then transfers them to the client who completes the final computation. This segmentation allows parallel processing and reduces the computational burden on either party, thereby decreasing overall computational time while maintaining security through the distributed nature of the computation.
Solution Approach 2:
Random values act as intermediaries in the computation process. The server generates these random values and uses them to compute intermediate results that are transferred to the client. These intermediaries enable the secure calculation of Hamming distances without requiring either party to directly process or reveal their actual data, thus preserving confidentiality while accelerating computation through efficient intermediate result exchange.
2Reliability
If traditional secure computation methods are used, then security is maintained, but the system cannot efficiently determine multiple Hamming distances in parallel
Solution Approach 1:
The computation is segmented into independent server-side operations (generating random values, computing intermediate results) and client-side operations (receiving intermediaries, computing final results). This segmentation enables multiple Hamming distance calculations to be performed in parallel by distributing different intermediate results to multiple clients simultaneously, thereby improving productivity while maintaining security through the segmented architecture.
Solution Approach 2:
The server's intermediate result computation serves multiple functions: it can be used to calculate Hamming distances against multiple different client codes simultaneously. The same intermediate results structure supports parallel comparisons across multiple data pairs, making the system universal and highly productive for batch biometric identification tasks while preserving security constraints.
3Loss of information
If secure computation protocols are implemented, then data privacy is protected, but substantial computing power and resources are required
Solution Approach 1:
Computational resources are segmented and distributed between server and client based on their respective capabilities. The server performs resource-intensive operations of generating random values and computing intermediate results, while the client performs lighter operations of receiving intermediaries and computing final results. This segmentation optimizes resource utilization across the system, reducing overall energy consumption while maintaining strong data privacy protection through the distributed secure computation framework.
Solution Approach 2:
Random values and intermediate results serve as computational mediaries that reduce the overall resource requirements. By pre-computing intermediate results using random values on the server side, the system avoids the need for both parties to perform full secure computation independently. This intermediary approach significantly reduces total computational resources and energy consumption while maintaining data privacy through the mathematical properties of the intermediate representations.
Data Source
AI summary
The invention relates to a secure method of processing data in which method is implemented the evaluation of a function that may be written as a linear combination of sub-functions with two binary inputs, in which a client and a server each possess a binary code, comprising n indexed bits, the method comprising the evaluation of the function with the binary codes of the client and of the server as inputs, without one of the client or the server obtaining information about the code of the other, the method being characterized in that it comprises the following steps: —the server randomly generates n indexed values and calculates the linear combination of these values with the same linear combination as that applied to the sub-functions to obtain the function, —the client implements, for each bit of his binary code, a technique of unconscious transfer to obtain from the server an intermediate data item comprising the randomly generated value of same index as the bit of the code of the client, increased by the value of the corresponding sub-function evaluated at the bit of same index of the code of the server and at said bit of his binary code, and —the client performs a linear combination of the intermediate data for all the bits of his binary code, with the same linear combination as that applied to the sub-functions to obtain the function, in such a way as to obtain a final result comprising the linear combination of the randomly generated values, increased by the evaluation of the function at the two binary codes.