Optimized verifiable delegated private set intersection on outsourced private datasets

Abstract

Private Set Intersection (PSI) has been applied in various fields, such as human genome research, advertising conversion rate analysis, etc. Traditional PSI has the drawback of requiring local devices to be constantly online and needing high storage capacity of the devices. To overcome these issues, many researchers have focused on delegating PSI computation to cloud servers. However, third-party clouds may bring many new challenges. One of the primary concerns is ensuring that the cloud server does not obtain sensitive data when the client outsources their dataset. Additionally, ensuring clients verify the correctness of the results is also a critical issue. In this paper, we propose the optimized verifiable delegated private set intersection protocol on outsourced private datasets (VO-PSI). Our protocol enables clients to verify the correctness of intersection with a negligible probability of incorrect results while being able to guarantee data privacy. Compared with previous research, the verifiability of the protocol has been dramatically improved. We analyze the security of the protocol under the malicious model and conduct experiments to evaluate its efficiency and feasibility.