Collaborative Research: CNS Core: Medium: Inference and Control in Overlay Networks

Description

An overlay network is a layered network, where the overlay layer makes use of the underlay layer in order to deliver data to its destinations. Overlay networks arise in many application scenarios such as content distribution networks that are used to distribute content over the Internet (e.g., movies, music, etc.), and are often used to deploy new technology that is incompatible with legacy devices and protocols. A key challenge in such systems is that the overlay cannot observe the innerworkings of the underlay, making it difficult to use the underlay efficiently. This project will develop mechanisms to learn the network topology and congestion of the underlay, and network algorithms for routing messages efficiently across overlay networks.

Existing overlay systems mostly rely on simple models of the underlay network, and may fail to achieve good performance when the underlay nodes cannot be fully observed and controlled. This project will develop fundamental limits and practical algorithms for monitoring and controlling partially observable/controllable overlay-underlay networks. This will be accomplished through two interdependent thrusts: thrust 1 will develop techniques that utilize measurements and side information observable to the overlay in order to infer the underlay network structure and state; and thrust 2 will develop algorithms that utilize the inferred information to control the operation of overlay nodes so as to optimize the performance for overlay services.

This collaborative project between Massachusetts Institute of Technology (MIT) and The Pennsylvania State University (PSU) brings together expertise on network inference and network control to tackle the problem of joint inference and control in overlay networks. The results of this project will enable a more efficient migration to next-generation networks and will facilitate system manageability for increasingly complex networks. The project will also support educational activities at the participating institutions and help broadening participation in computing of under-represented groups through summer programs at MIT and PSU.

Participants

PI: Ting He (Penn State)

PI: Eytan Modiano (MIT)

PhD students: Yudi Huang, Tingyang Sun (Penn State), Sathwik Chadaga (MIT)

MS students: Lay Patel (Penn State)

Significant Results:

• Optimized cross-path attacks (lead: PSU): consistent estimation of link sharing between actively probed paths and passively monitored paths
• Underlay-aware overlay routing (lead: PSU): detection of links shared between overlay tunnels and estimation of effective link capacities, both from overlay measurements
• Overlay/underlay scheduling (lead: MIT): universal tracking algorithm that can mimic any scheduling algorithm under delayed observations, with applications in NextG wireless networks
• Network utility maximization in overlay-underlay networks (lead: MIT): tracking drift-plus-penalty algorithm that operates only on overlay nodes with an uncontrollable and unobservable underlay

Publications:

Journal

1. Yudi Huang and Ting He, Overlay Routing Over an Uncooperative Underlay, accepted to IEEE/ACM Transactions on Networking, March 2025. [Supplementary Materials]
2. Bai Liu, Quang Minh Nguyen, Qingkai Liang, and Eytan Modiano, “Tracking Drift-Plus-Penalty: Utility Maximization for Partially Observable and Controllable Networks,” IEEE/ACM Transactions on Networking, vol. 32, no. 2, pp. 1064 – 1079, April 2024.
3. Yudi Huang, Yilei Lin, and Ting He, Optimized Cross-Path Attacks via Adversarial Reconnaissance, Proceedings of the ACM on Measurement and Analysis of Computing Systems, vol. 7, no. 3, article 58, December 2023.
4. Tian Xie, Sanchal Thakkar, Ting He, Patrick McDaniel, and Quinn Burke, Joint Caching and Routing in Cache Networks with Arbitrary Topology, IEEE Transactions on Parallel and Distributed Systems, vol. 34, no. 8, pp. 2237-2250, August 2023.

Conference

1. Jinyi Yoon, Jiho Lee, Ting He, Nakjung Choi and Bo Ji, S2M3: Split-and-Share Multi-Modal Models for Distributed Multi-Task Inference on the Edge, IEEE ICDCS, July 2025.
2. Sathwik P Chadaga (MIT), Eytan Modiano (MIT), Drift Plus Optimistic Penalty – A Learning Framework for Stochastic Network Optimization, IEEE INFOCOM, May 2025.
3. Yudi Huang, Tingyang Sun, and Ting He, Overlay-based Decentralized Federated Learning in Bandwidth-limited Networks, ACM MobiHoc, October 2024.
4. Yudi Huang, Yilei Lin, and Ting He, Optimized Cross-Path Attacks via Adversarial Reconnaissance, ACM Sigmetrics, June 2024.
5. Tian Xie, Sanchal Thakkar, Ting He, Novella Bartolini, and Patrick McDaniel, Host-based Flow Table Size Inference in Multi-hop SDN, IEEE Globecom, December 2023.
6. Yudi Huang and Ting He, Overlay Routing over an Uncooperative Underlay, ACM MobiHoc, October 2023. [Supplementary material]
7. Bai Liu and Eytan Modiano, “Universal Policy Tracking: Scheduling for Wireless Networks with Delayed Observation,” Allerton Conference, September 2022.

Tutorial/Seminar/Invited Talks:

1. “Actionable Network Tomography: Topology Inference and Applications“, Seminar at the Laboratory for Information and Decision Systems (LIDS), MIT, 2025.
2. “Rethinking Network Monitoring in Layered Networks”, Keynote at IEEE International Conference on Network Softwarization, 2024.
3. “A Comprehensive Overview of Network Tomography: Inverse Methods for Network State Monitoring from End-to-End Measurements”, Tutorial at ACM SIGMETRICS (jointly with IFIP Performance), 2024.
4. “CT Scan for Your Network: Topology Inference from End-to-End Measurements”, Joint Seminar for NSF AI-EDGE Institute and OSU ECE Department, Ohio State University, 2024.
5. “CT Scan for Your Network: Topology Inference from End-to-End Measurements”, Center for Networked Intelligence (CNI) Seminar, IISc, 2023.
6. “When Networking Meets Learning: Monitoring, Security, and Support of Distributed Computing”, Penn State-IISc Joint Workshop, 2023.