Secure Multi-Party Private Set Intersection with Semi-Honest Nodes

In this paper we consider a multi-party private set intersection system consisting of N source nodes, each has access to a private set, and a master who has access to a public set $\mathcal{A}$. In this system we assume that up to T of the sources are semi-honest, i.e., they follow the protocol but they may share their private data with each other to gain some information about the private set of the other nodes. The master is interested in finding the elements of $\mathcal{A}$ which appeared in at least K of the private set. First, the master sends the public set to the source nodes, and then each source sends a randomized function of its private set and the public set to each of the other sources, called its share. After that, the nodes start communicating with each other and processing their shares, and finally send some results to the master. In this paper, we propose a secure private set intersection scheme that orderwise reduces the communication and computation complexity compared to the conventional information-theoretic private set intersection schemes.