Xu (Eric) Wang

I am Xu Wang (Eric Wang, 王旭), a Ph.D. candidate in the School of Computing at Queen’s University (WineMocol Lab), co-supervised by Prof. Yuanzhu Chen (Queen’s University) and Prof. Octavia Dobre (Memorial University).

My research centers on decentralized federated learning (DFL) over real-world networks, where devices train a shared model with their neighbors and no central server. I study how the topology of the underlying network shapes what such a system can learn, and how to make it robust. On the design side, I introduced teleportation links (a few long-range links added to the network) and connectivity enrichment to mitigate catastrophic forgetting and community isolation in DFL networks. On the security side, I study community-level attacks that disrupt cross-community information flow, as well as defenses against malicious models in federated learning deployed on real-world energy storage systems.

I collaborate with researchers at Memorial University, the University of Calgary, and Korea Maritime and Ocean University. See the full publication list or my Google Scholar profile.

Contact: xu.wang@queensu.ca · School of Computing, Queen’s University, Kingston, Ontario, Canada

At a Glance

• 15 peer-reviewed publications (13 as first author), including two first-author articles in IEEE Transactions on Network Science and Engineering (2026) — Publications
• Instructor for CISC 365 (Algorithms; 103 students) at Queen’s University and for three offerings of CISC 102 (Discrete Structures; 116, 146 and 418 students) — Teaching
• TPC Member for IEEE ICC 2026 and 2027, IEEE CCECE 2026, WCSP 2026, and IEEE VTC2024-Fall; Session Chair at IEEE ICC 2024 and 2025 — CV

Selected Publications

1. DOCAttack: Data-Oriented Community Attack in Decentralized Federated Learning. X. Wang, Y. Chen, Q. Ye, J. Son, and O. A. Dobre. IEEE Transactions on Network Science and Engineering, 2026. Fuses leaked client data profiles with network structure to find data-oriented communities. Under a given budget of node and edge removals, it then isolates communities that topology-only attacks cannot separate. [Paper]
2. Teleportation Links: Mitigating Catastrophic Forgetting in Decentralized Federated Learning. X. Wang, Y. Chen, Q. Ye, and O. A. Dobre. IEEE Transactions on Network Science and Engineering, 2026. Shows how non-IID bias spreads over a DFL topology and why nodes forget; a few long-range teleportation links speed up knowledge sharing with no central server. [Paper]
3. Federated Learning for Anomaly Detection: A Case of Real-World Energy Storage Deployment. X. Wang, Y. Chen, and O. A. Dobre. IEEE International Conference on Communications (ICC), 2022. Trains one global autoencoder across battery sites with federated learning and flags faulty batteries right after commissioning, with no raw data leaving a site. [Paper]

Research Interests

• Federated Learning and Distributed Machine Learning
• Computer Networks and Network Science
• Cyber Security

News

• September 2026: Instructor for CISC 365 (Algorithms) at Queen’s University, Fall 2026, co-taught with Prof. Yuanzhu Chen (103 students).
• June 2026: Our paper “DOCAttack: Data-Oriented Community Attack in Decentralized Federated Learning” was accepted by IEEE Transactions on Network Science and Engineering.
• May 2026: Presented “Disrupting Cross-Community Information Flow in Decentralized Federated Learning” at the IEEE INFOCOM 2026 Workshops in Tokyo, Japan.

Education

• Ph.D. in Computing (expected November 2026), Queen’s University, Kingston, Ontario, Canada; transferred from Memorial University (2020–2021)

• M.Sc. in Signal Processing and Intelligent Control (2018), Inner Mongolia University of Technology, China

• B.Sc. in Communication Engineering (2015), Shenyang Institute of Technology, China

Awards and Recognition

• Graduate Research Award (Second Place), IEEE Kingston Section, 2025
• Student Travel Award, Queen’s University, 2024, 2025
• Student Travel Award, Canadian Artificial Intelligence Association, 2023
• Mitacs Accelerate Fellowship, 2023
• Queen’s Graduate Award, 2021–2024
• National Scholarship for Postgraduates, Ministry of Education of China, 2018
• Outstanding Graduate, Inner Mongolia University of Technology, 2018
• Merit Student, Inner Mongolia University of Technology, 2017

Academic Service

Professional Service:

• TPC Member, IEEE ICC 2026, 2027
• TPC Member, IEEE CCECE 2026
• TPC Member, WCSP 2026
• Session Chair, IEEE ICC 2025
• TPC Member, IEEE VTC2024-Fall
• Session Chair, IEEE ICC 2024
• Graduate Mentor, Google exploreCSR (Queen’s University), 2022
• Student Volunteer, IEEE ICC 2025

Departmental Service:

• Student Representative, Appointments Committee, School of Computing, Queen’s University (2023–present)
• Student Representative, Renewal, Tenure, and Promotion Committee, School of Computing, Queen’s University (2023–present)

Reviewer for 12 journals (including IEEE JSAC, IEEE Internet of Things Journal, IEEE TNSE, and ACM Computing Surveys) and 10 conferences and symposia (including IEEE ICC, Globecom, and WCNC).

Research & Teaching Impact

My research and teaching align with the United Nations Sustainable Development Goals:

• SDG 4 (Quality Education): Across four offerings at Queen’s University I have taught Discrete Structures and Algorithms, the mathematical foundations a computing degree rests on, to 783 students. Tiered learning paths and assessment choice let students who arrive with uneven preparation reach the same standard, and as a Google exploreCSR graduate mentor I brought undergraduates into computing research.

• SDG 7 (Affordable and Clean Energy): Battery storage is what turns intermittent renewable generation into a supply you can count on, and it degrades quietly. Across four papers on energy storage systems, one of them a real-world deployment, I built federated learning that flags faulty batteries soon after commissioning and holds up when some sites contribute malicious model updates, without raw operating data ever leaving a site.

• SDG 9 (Industry, Innovation and Infrastructure): Decentralized federated learning drops the central server, so devices can learn together over whatever network they already have. I design the topologies that make this practical, using teleportation links and connectivity enrichment so that isolated communities are not left behind, and I study the community-level attacks such networks have to survive, work that carries over to mesh networks and 6G.