Causal and decision-aware management for integrated communication, transportation, and building energy systems

Modern energy-consuming systems are becoming increasingly interconnected. Future 5G/6G base stations, electric-vehicle charging services, renewable generation, battery storage, and building energy systems will form cyber-physical energy systems in which demand, mobility behavior, weather, prices, and control decisions jointly determine cost, reliability, and sustainability. This thesis, for the first time, proposes the energy-communication-transportation hub (ECT-Hub) concept, centered on base stations that integrate communication load, electric-vehicle charging demand, renewable generation, battery storage, dynamic electricity prices, and carbon-aware operation into one decision system.  In this thesis, we study how such converging energy systems should choose effective interventions, execute them over time, and forecast their consequences under changing environments. Building on this system abstraction, the thesis develops a unified causal decision framework in three layers. First, ECT-Price treats charging discounts as interventions and uses causal inference to identify when incentives are likely to increase electric-vehicle charging demand, avoiding unnecessary discounts for users who would charge regardless. Second, ECT-DT uses Decision Transformer-based sequence modeling to schedule batteries under shifting demand, renewable generation, and prices. Third, Causal EnergyTwin extends the framework to intervention-aware time series forecasting with multi-site building and industrial energy measurements. Its measurement study shows climate shift, schedule dependence, policy-weighted action potential, and degraded out-of-distribution forecasting, motivating counterfactual forecasts under weather, calendar, tariff, and carbon context. Together, the three layers decide when interventions matter, how to act on them, and what consequences they may produce. Extensive evaluations using real-world data demonstrate the effectiveness and potential of the proposed framework over state-of-the-art solutions.