Northwestern Networks Group Team Wins a Best Student Paper Award at EuroSys 2024

A Northwestern Networks Group team, including computer science PhD students Yunming Xiao and Sen Lin, earned the Best Student Paper Award last month at the EuroSys 2024 conference

A Northwestern Networks Group team led by Yunming Xiao, a PhD candidate in computer science at Northwestern Engineering, earned a Best Student Paper award at EuroSys 2024, held April 22 – 25 in Athens, Greece.

EuroSys is a premier systems software research and development conference focused on topics including operating systems, database systems, real-time systems, networked systems, storage systems, middleware, distributed, parallel, and embedded computing systems.

Aleksandar Kuzmanovic (l) and Yunming Xiao (r)Xiao was first author of the winning paper “Snatch: Online Streaming Analytics at the Network Edge,” co-authored by computer science PhD student Sen Lin; former visiting undergraduate research assistant Yibo Zhao; and Aleksandar Kuzmanovic, professor of computer science at the McCormick School of Engineering.

In their paper, the team posed the question, “What if internet cookies stored all user features for analytics instead of only identifiers such as user IDs?”

The paper explores the potential of catching and pre-processing user clicks before they reach the server’s data centers. The group aimed to develop a system to enable early click catching, in-network processing, and anonymity-preserving analytics, and to quantify the achievable performance benefits.

They designed and implemented ‘Snatch,’ an analytics system that early forwards and pre-processes online streaming data at the network edge to speed up the online streaming analytics and preserve user privacy and security. The approach utilizes ‘semantic cookies,’ or carriers of encrypted user information that is individually unidentifiable and directly available for analytics.

By leveraging existing edge infrastructures, Snatch preserves user anonymity during web browsing and improves the performance of online streaming analytics.

“Essentially, by using Snatch and semantic cookies, the web providers will no longer be able to trace each individual but can only receive aggregated information about the user group,” Xiao said.

In addition to enhancing user privacy and optimizing online streaming analytics, Xiao noted additional potential benefits of semantic cookies, such as improving personalized online advertising.

Xiao is now investigating how to enhance the privacy of other internet components such as domain name service (DNS), a fundamental service behind every web request that translates domain names, such as "northwestern.edu," to machine-understandable IP addresses.

McCormick News Article