Privacy Set Intersection Protocol based on Homomorphic Encryption and Optimization Filters

Privacy Set Intersection Protocol based on Homomorphic Encryption and Optimization Filters

연구 분야: Cryptography

학회: CNSSE '25: Proceedings of the 2025 5th International Conference on Computer Network Security and Software Engineering

With the increasing demand for privacy protection, Privacy Set Intersection (PSI), as a key cryptographic protocol in Secure Multi-Party Computing (MPC), has gained significant applications in several fields. Existing protocols have made significant progress in terms of time overhead and communication overhead, but there is still room for optimization when facing large-scale, unbalanced data scenarios. In this paper, we propose a two-party privacy set intersection protocol based on semi-homomorphic encryption from the perspective of privacy protection to address the communication overhead problem and ensure the security of the protocol. In order to reduce the time overhead, this paper adopts the preprocessing technique based on optimized filtering to narrow down the set of intersection candidates, and at the same time, optimizes the process of comparing the set elements by using hash substitution sorting to further improve the time efficiency of the protocol, and combines with the Oblivious Polynomial Evaluation (OPE) technique to ensure the data privacy. Experimental results show that the protocol is able to compute the intersection of private sets securely under the semi-honest model. Compared with the CM20 protocol, when the collection size of one party is 100,000, the other party is 100, and the length of the data item is 280bit, the protocol time overhead is reduced by 30.08%, and the communication overhead is reduced by 14.53%, which is more practical for unbalanced scenarios with large data size. The protocol proposed in this paper provides new ideas and methods for efficiently realizing privacy set intersection, which is of great significance for promoting the development of privacy computing.