Please note: This master’s thesis presentation will take place online.
Onur Eren Arpaci, Master’s candidate
David R. Cheriton School of Computer Science
Supervisors: Professors Florian Kerschbaum & Raouf Boutaba
The Over Threshold Multiparty Private Set Intersection (OT-MP-PSI) is a cryptographic problem where N participants aim to identify elements that appear in at least t sets without revealing information about elements that do not meet this threshold. This protocol is particularly relevant for detecting multi-institution attacks, which require private and efficient information sharing. In order to detect the attackers, the institutions want to identify the common external IPs among other institutions without leaking the rest of the connection data. Existing solutions either require numerous communication rounds or have impractical computational complexities for large datasets. We introduce a non-interactive OT-MP-PSI protocol with a computational complexity of O(t^2 M C(N,t)), optimized by a novel hashing scheme.
Experimental benchmarks using real-world network data demonstrate the protocol's practicality, showing significantly faster reconstruction times compared to existing methods and effective scalability with the number of participants and set sizes.
Our protocol also provides optimal performance for the two-party private set intersection (N=t=2) and the multi-party private set intersection (N>2, t=N), offering a versatile solution for various applications.
You can attend this presentation on Zoom.