Predictive Orchestration and Adaptive Forecasting

This research area focuses on forecasting that is useful because it changes decisions, not simply because it reduces prediction error. The goal is to connect estimates of future traffic, congestion, energy demand, thermal behaviour, and service quality directly to orchestration actions across radio, compute, edge, cloud, and sensing resources.

A central theme is adaptive intelligence under changing conditions. That includes online model selection, uncertainty-aware forecasting, and control systems that can decide when better prediction is worth the computational cost and when lighter-weight action is sufficient. The broader aim is to move from reactive operations toward anticipatory, decision-aware AI-native systems.