A Signal‐Matching Approach to RLWE‐Based PSI: Achieving Optimal Online Efficiency in 3 Rounds
Yue Qin, Bei Liang, Hongyuan Cai +2 more
In this study, we propose a novel post‐quantum private set intersection (PSI) protocol based on the ring learning with errors (RLWE) assumption. Different from existing PSI protocols based on post‐quantum oblivious pseudorandom functions (EUROCRYPT’25) and fully homomorphic encryption (FHE) (CCS’21), our construction follows the Diffie–Hellman (DH)‐based PSI paradigm and uses post‐quantum key exchange to achieve PSI functionality. Our key technique is a signal matching (SigMa) mechanism, which determines whether two items match by checking the consistency of the reconciliation signals derived from their corresponding shared keys. The resulting PSI protocol completes in three rounds. In the first two rounds, the parties exchange RLWE‐based public keys associated with their items; in the third round, the sender transmits the corresponding signals. The receiver then applies SigMa to compare the received signals with its locally computed signals and thereby identify the intersection. We implement our protocol and compare it with Cong et al. (CCS’21) and Heimberger et al. (EUROCRYPT’25). Our protocol achieves the lowest online communication across all evaluated symmetric and asymmetric set‐size configurations, at the cost of additional offline communication. It also achieves the lowest online and total times among the evaluated protocols for all symmetric settings with | X | = | Y | ∈ {2 12 , 2 16 , 2 20 } and all asymmetric settings with | X | ∈ {2 12 , 2 16 , 2 20 } and | Y | ∈ {5535, 11041}. We believe that our work not only supplements existing post‐quantum PSI frameworks but also opens the opportunity for broader applications of post‐quantum PSI.
IET Information Security•Jan 1, 2026•0 citations