AsTAR: Sustainable Battery Free Energy Harvesting for Heterogeneous Platforms and Dynamic Environments

Today’s commercial Internet of Things devices remain largely dependent upon batteries, which offer high capacity, stable energy storage at the expense of limited shelf-lives and toxic chemical compositions. Research on sustainable energy harvesting platforms is essential to realizing a new generation of long-lived and environmentally friendly IoT products. This paper contributes to this goal by introducing AsTAR, an energy-aware task scheduler and associated reference platform that aims to lower the burden of developing sustainable applications through self-adaptive task scheduling. We evaluate AsTAR based on its capability to deliver sustainable operation on heterogeneous platforms. Evaluation shows that: (i.) With zero modeling AsTAR rapidly identifies optimum task scheduling rates, while (ii.) reacting quickly to environmental change and (iii.) these features incur minimal performance overhead in terms of memory, computation and energy. Considered in sum, we believe that these features significantly simplify the process of creating sustainable energy harvesting IoT applications.