Research
My group develops the theory and systems for timely, economically sound, and privacy-preserving intelligent networks — spanning the age of information, network economics, and trustworthy learning at the network edge. Representative work is grouped into three themes below.
Timeliness & Edge Intelligence
Delivering fresh information and low-latency intelligence at the network edge: age-of-information-optimal scheduling, mobile-edge computing, and collaborative inference across edge large language models.
The MoE² framework routes each prompt across a network of edge LLM experts via a gating and two-level expert-selection mechanism. — MoE² (IEEE/ACM ToN, 2026)
- Timely CPU Scheduling for Computation-intensive Status Updates — IEEE Transactions on Information Theory, 2026
- MoE²: Optimizing Collaborative Inference for Edge Large Language Models — IEEE/ACM Transactions on Networking, 2026
- Asynchronous Fractional Multi-Agent Deep Reinforcement Learning for Age-Minimal Mobile Edge Computing — IEEE/ACM Transactions on Networking, 2026
- Age of Information in Mobile Networks: Fundamental Limits and Tradeoffs — ACM MobiHoc, 2024
Network Economics & Mechanism Design
Pricing, incentives, and mechanism design for networks and data markets: how to price fresh data, elicit truthful participation, and design efficient markets under private information.
Fresh-data markets connect information sources and users across IoT, real-time cloud, big-data, and financial applications. — Pricing Fresh Data (IEEE JSAC, 2021)
- Large-Scale Mechanism Design for Networks: Superimposability and Dynamic Implementation — IEEE Transactions on Mobile Computing, 2025
- Optimal Mechanism Design for Heterogeneous Client Sampling in Federated Learning — IEEE Transactions on Mobile Computing, 2024
- Pricing Fresh Data — IEEE Journal on Selected Areas in Communications, 2021
- Truthful Mobile Crowd Sensing with Interdependent Valuations — ACM MobiHoc, 2020
Trustworthy & Privacy-Preserving Learning
Provable privacy and integrity for learning systems: age-dependent differential privacy, certified machine unlearning, and faithful (incentive-compatible) federated learning.
Age-dependent differential privacy: aging and noise injection protect privacy-sensitive fresh data against untrusted clients. — Age-Dependent Differential Privacy (IEEE TIT, 2024)
- Age-Dependent Differential Privacy — IEEE Transactions on Information Theory, 2024
- Hessian-Free Online Certified Unlearning — International Conference on Learning Representations (ICLR), 2025
- Unlearning Incentivizes Learning under Privacy Risk — ACM Web Conference (WWW), 2025
- Faithful Edge Federated Learning: Scalability and Privacy — IEEE Journal on Selected Areas in Communications, 2022