.png){kind=icon}
.png){kind=icon}
Return
Scalable multiparty threshold private set intersection based on additive secret sharing
.png){kind=icon}
PRE
AI
DOI：10.1016/j.jisa.2026.104598.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
MTPSI is a special case of secure multiparty computation, where each party has a set and computes the intersection of these sets if the intersection cardinality is greater than a threshold. If the intersection cardinality is less than the threshold, the parties do not get any information about it. Although the TPSI protocol has many real-world applications, such as online carpooling, contact discovery, machine learning, etc., most of the current research concentrates on two-party setting and pay less attention to multi-party setting. In addition, the existing TPSI protocols have the problems of high computational overhead, leaking intersection cardinality, inefficiency in running on large sets (e.g., 220) and inability to scale to more parties (e.g., 15 parties). In order to solve these problems, this paper proposes a new Multiparty Private Membership Test (MPMT) protocol based on multi-point oblivious PRF, oblivious key-value stores, Cuckoo hash, and additive secret sharing. Then, we implement MTPSI protocol using Vector Oblivious Match Test (VOMT) and the MPMT protocol, which outperforms previous work in runtime and proves its security under a semi-honest model. We implemented our protocol in C++ and evaluated it under different settings of set size and the number of parties. Our experimental results demonstrate that the running time of our protocol is 0.766 seconds when the set size is 211 and 2 parties (the threshold t = 0.5 × 211 ), whereas the previous state-of-the-art TPSI protocol takes 700 seconds. Additionally, our protocol has a running time of 268.788 seconds under the setting of 220 set size and 15 parties (the threshold t = 0.5 × 220 ). Our MTPSI protocol is a practical solution that can perform efficiently in scenarios with large datasets and multiple parties.
Keywords:
Multi-party computation
Threshold private set intersection
Oblivious key-value stores
Additive secret sharing
Journal
IF:
3.7
Papers:
1.9K
Citations:
4.9K