Efficient Server-Aided Multiparty Private Set Intersection

doi:10.1109/tcc.2026.3742776

Efficient Server-Aided Multiparty Private Set Intersection

2026-10-12

0

PRE

AI

DOI：10.1109/tcc.2026.3742776

Original

Original request for help

Save

Abstract

En 中文

Multiparty private set intersection (MPSI) enables parties to compute the intersection of private datasets without revealing extra information. The performance can be enhanced by outsourcing participants’ data to an untrusted cloud-a technique known as server-aided MPSI-which significantly reduces computation and communication overhead for participants. In this paper, firstly we analyze the correctness of two newly proposed server-aided MPSI protocols, namely MDPPC (PST 2023) and DMPSI (IEEE Trans. on Services Computing 2024), the core building block of which is a two-dimensional hash table obtained by applying simple hash twice to the set items and used for efficient aggregation. We show that their approach fails to insert items to a one-to-one position of the two-dimensional hash table with high probability, leading to the failure of protocol execution. Secondly, we propose a server-aided MPSI protocol based on OPRF and Bloom filter, enabling aggregation through cheap bitwise-AND operation. Our protocol is secure against semihonest adversaries allowing any number of colluding parties (at least one honest party), and achieves linear complexity in set size and independence from the number of parties on the leader side. We compare the performance of our protocol with stateof- the-art MPSI protocols. Experimental results demonstrate that our protocol achieves the lowest total communication and computation costs, as well as the lowest communication and computation costs of the leader, particularly in large-scale multiparty scenarios.

Keywords:

Multiparty private set intersection

server-aided

semi-honest secure

Efficient Server-Aided Multiparty Private Set Intersection

Abstract

Journal

Organization

Cited Papers