Our research focuses on integrated observation, prediction, control, and security for networked and cyber-physical systems, aiming to build efficient and robust information infrastructures that operate reliably in dynamic and uncertain real-world environments.
We develop methods to observe and analyze network traffic and states, predict future variations, and control networks in a stable manner. Integration of external sensor and real-world data enhances decision accuracy in network control.
In disaggregated data centers and edge–cloud environments, we study allocation and performance control of both communication and computation resources to achieve efficient and high-performance systems.
We investigate anomaly detection and prediction using IoT and multimodal sensor data, as well as robustness and defense strategies for machine learning models against sensor-based adversarial threats.
Our work includes development of probabilistic models and digital twin frameworks for real-world information integration and prediction, and control approaches that consider human–environment interactions.