Nets: RUI: privacy by design: enabling strong privacy guarantees with open-access network data

The National Science Foundation Division of Information and Intelligent Systems awarded Swarthmore College $248,362 on June 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop privacy-preserving techniques for sharing network data.

The project designs and analyzes differentially private mechanisms across three network data sharing scenarios. The first thrust develops methods for single network providers to share fine-grained workload traces with edge computing services while protecting individual users. The second thrust creates approaches for releasing sequential mobility traces, such as device movements across campus wireless access points, using machine learning models combined with formally calibrated noise. The third thrust addresses collaborative settings in which multiple Internet service providers may jointly compute statistics, including heavy hitter IP prefixes and intersection counts, without revealing their own raw traffic or identifiers. The work spans end-to-end algorithm development, formal privacy guarantees, and empirical evaluations on real network datasets. Broader impacts include education integration, open science contributions, and technology transfer to the networking community.

Performance will occur in Swarthmore, Pennsylvania, with a period of performance from June 1, 2026, through May 31, 2029. This is a Project Grant, a type of assistance funding for investigator-initiated research.

Abstract

Important decisions about the internet depend on data about the networks that help connect to everyday lives. However, these data are difficult to share privately and securely. Network and service providers are now able to collect records of how devices connect to wireless networks and how internet traffic flows through their systems. These records can reveal how services are used by users. Sharing such data would help researchers and engineers make networks faster, more reliable, and fairer, but releasing them directly would put individual privacy at risk. This project develops mathematically rigorous privacy preserving techniques, so that network operators can share useful versions of their data without exposing sensitive information about any single person or organization. The project designs and analyzes a suite of differentially private mechanisms tailored to three canonical network data sharing scenarios. The first thrust focuses on single network providers that wish to share fine-grained workload traces with edge computing services while protecting individual users. The second thrust develops methods for releasing sequential mobility traces, such as device movements across campus wireless access points, using machine learning models combined with formally calibrated noise. The third thrust targets collaborative settings in which multiple internet service providers may jointly compute statistics, such as heavy hitter IP prefixes and intersection counts, without revealing their own raw traffic or identifiers. Across these thrusts, the project plans to develop end-to-end algorithms, to address formal privacy guarantees, and to conduct empirical evaluations on real network datasets. The broader impacts of this project span education, open science, and technology transfer to the networking community. The work integrates privacy-preserving data analysis into undergraduate courses and year-long research experiences. The project releases open-source code, synthetic datasets, and teaching materials so that other instructors, researchers, and practitioners can learn how to apply differential privacy to network data. Outreach activities connect these ideas to kindergarten through grade 12 students and local community programs, illustrating how data-driven technologies intersect with questions of privacy and fairness. By demonstrating concrete, deployable methods for sharing network data safely, this project aims to lower barriers to collaboration between industry, academia, and policymakers on challenges in internet measurement and security. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are not planned for this award.