Resilient Scheduling of Energy-Variable Weakly-Hard Real-Time Systems

Cyber-physical systems and the like have accelerated the growth of demands for long-life energy-limited devices, encouraging the major trend of using energy harvesting from renewable sources like solar and wind. The intermittency of these energies enforces such energy-variable systems to possibly anticipate the changes, and efficiently adapt in face of either predicted or non-predicated changes, i.e. they must be energy-resilient. In this paper, we target weakly-hard real-time energy-variable systems with multiple performance levels. We formalize the concept of energy-resilience in such systems, define and prove some properties of the resilient systems, and give a scheduler with the aim of maximizing the system resilience through adaptive control of the system performance with respect to the energy changes. Also, we propose an online schedulability test that considers the initial available energy in the storage unit and the (m, k)-firm constraints. Furthermore, we give an energy-resilient scheduling algorithm which employs model predictive control. The simulation results show that our proposed method performs very good in approximating the optimal solution.