Research

Self-Driving Networks

The long-term research goal for Systems and Networking Lab (SNL) is to develop self-driving networks that leverage advances in different research areas (i.e., networking, machine learning, security, and computer architecture) to address the digital inequity issues. Specifically, our goal is to build self-driving last-mile networks (e.g., cable, mobile, satellite, etc.) that can lower the cost of operating highly available, reliable, performant, and secure network connectivity, highly desired for under-served communities.

To this end, our group explores how we can (1) enable accurate and flexible (streaming) analytics over network data at scale; (2) lower the threshold to curate high-quality datasets for different learning problems from diverse network environments at scale; (3) develop generalizable and robust learning models that can both accurately assess the network’s state and take effective actions to keep networks performant and secure; and (4) establish trust in ML-based artifacts so network operators feel confident enough to relinquish control to these artifacts in production settings.

Related Publications

Some of the recently published works that provide a sample of some of the ongoing activities at SNL:

• AI/ML for Network Security: The Emperor has no Clothes, ACM CCS, 2022.
• The Importance of Contextualization of Crowdsourced Active Speed Test Measurements, ACM IMC, 2022.
• Detecting Ephemeral Optical Events with OpTel, USENIX NSDI, 2022.
• Internet Inequity in Chicago: Adoption, Affordability, and Availability, TPRC, 2022.
• DynamiQ: Planning for Dynamics in Network Streaming Analytics Systems, arXiv: Report 2106.05420, 2021.
• An Effort to Democratize Networking Research in the Era of AI/ML, ACM HotNets 2019.

Funding

The research in my group is funded by National Science Foundation (NSF) and different network service providers (Verizon Innovations, ViaSat) and vendors (Intel).

You can find more details about some of the funded projects here:

• IMR: RI-P: Programmable Closed-loop Measurement Platform for Last-Mile Networks (NSF)
• IMR: MM-1A: ADDRESS: Augment, Denoise and Debias Crowdsourced Measurements for Statistical Synthesis of Internet Access Characterization (NSF)
• CC* Integration-Large: Democratizing Networking Research in the Era of AI/ML (NSF)
• CC* Integration-Large: Bringing Code to Data: A Collaborative Approach to Democratizing Internet Data Science (NSF)
• The Estimation and Monitoring of Quality of Experience Delivered over Internet Services (ViaSat)
• MLWiNS: RL-based Self-driving Wireless Network Management System for QoE Optimization (NSF and Intel)
• Scaling Cybersecurity Infrastructure using Programmable Data Planes (Verizon)