How can a connected device operate when it never knows exactly how much energy it will have available?

From environmental sensors to medical implants, an increasing number of embedded systems can harvest energy from their surroundings, for example through solar power.

But this autonomy comes with a challenge: these devices need to complete their tasks within a given timeframe while avoiding consuming more energy than they have available.

In their latest study, published in the Journal of Systems Architecture, Ruizhe QIU (PhD at IDIA, Télécom Paris, Institut Polytechnique de Paris) and his co-authors introduce HELIOS, a new method designed to address both constraints simultaneously: time and energy.

How does it work?

The idea is to determine in advance when each task can be executed, based not only on its deadline but also on the available energy. A task only starts when the system has enough energy to guarantee that it can run to completion.

To evaluate this approach, the researchers simulated 70,000 task sets with varying levels of processor workload and energy availability.

The results show that HELIOS produces more valid execution schedules than the benchmark methods tested, particularly when energy constraints become more significant.

In the future, HELIOS could be tested on real-world systems and extended to scenarios that more closely reflect actual operating conditions, such as variable energy harvesting or multiprocessor systems.

To learn more, read the scientific paper: https://www.sciencedirect.com/science/article/pii/S1383762126002031?via%3Dihub

Co-authored by Thomas Robert, Samuel Tardieu, Laurent Pautet and Frank Singhoff.