1/15/2026 Lauren Laws
Assistant Professor Yupeng Zhang and his team, along with researchers from the University of California, Irvine, received a $260,000 grant from the National Science Foundation to develop a working program that would further increase the efficiency of intrusion detection systems, such as firewalls and antivirus software, while simultaneously preserving data privacy.
Written by Lauren Laws
Can you detect a burglar inside a house without seeing the house itself and its contents? You need to know what’s there before you can determine whether something is amiss. What if that burglar was a virus, and the house your computer and all your data? The question now becomes: Can antivirus software find the threat without seeing your personal data?
Assistant Professor Yupeng Zhang and his team, along with researchers from the University of California, Irvine, want to develop a working program that would further increase the efficiency of intrusion detection systems, such as firewalls and antivirus software, while simultaneously preserving data privacy. They received a $260,000 grant from the National Science Foundation for their project, “Practical Secure Multiparty Computations for Graph-based Intrusion Detection Systems.” The research is part of a special program from the NSF called Privacy Preserving Data Sharing in Practice (PDAPs). Current intrusion detection systems work by accessing a device’s data to detect threats. When you download or use antivirus software, you agree to share your data with the company that created the software.
“Imagine a case where a company is going to build a firewall for a larger network like an entire department instead of just my own computer. Information such as when you turn on your
computer, which website you’re visiting, and your communications is collected,” explained Zhang. “All of this information needs to be logged and analyzed to detect whether there is a malicious behavior or not.”
The project’s goal was to address these privacy issues by enabling privacy-preserving analytics for intrusion detection—without sharing users’ data with the company.
“I worked on something similar for so-called privacy preserving machine learning training in the early days. Developing cryptographic protocols where you only know the result of this machine learning model classification, but not anything about your sensitive data, model parameters, or intermediate results.”
For the project, Zhang will work on designing cryptographic protocols while lead PI Professor Zhou Li from the University of California, whose research focuses on cybersecurity on real- world applications, will create the firewall and cybersecurity in practice.
So far, the team has developed research components such as special protocols and techniques related to cryptography, as well as preliminary results and prototype implementations. They eventually plan to develop an end-to-end system with multi-layer architecture for the program to be user-friendly.
Grainger Engineering Affiliations
Yupeng Zhang is an Illinois Grainger Engineering assistant professor of electrical and computer
engineering in the Department of Electrical and Computer Engineering and the Siebel School of
Computer and Data Science. He is affiliated with the Coordinated Science Laboratory.