Age of Anomalies for Energy-Constrained Networked Systems

Kurzfassung

This paper investigates a finite-horizon transmission control problem for a warning-based cyber-physical system with three operating modes: stable, vulnerable, and critical. In this setting, a sensor monitors the system and communicates updates to a controller under a finite-capacity battery with energy harvesting. At each time slot, the sensor decides whether to remain idle, transmit a compressed update, or transmit a dense update, with different energy costs and different stabilization effectiveness. Our objective is to design a finite-horizon transmission control policy under the battery dynamics. Performance is measured by a state-aware metric that captures both the onset of the critical mode and the duration of remaining in it. We formulate the problem as a finite-horizon Markov decision process and solve it via dynamic programming. Numerical results illustrate a time-varying threshold-type structure in the optimal policy, quantify when simple mode-based heuristics are near-optimal, and highlight the roles of critical-mode persistence and battery capacity.