EDF Option for Scheduling Real-Time Jobs in Energy Harvesting Systems

The introduction of energy harvesting functionalities into an embedded system such as wireless sensor network raises many design questions. The central one is how to smartly use harvesting abilities so as to optimize performance and lifetime of the system? How to dynamically adapt the processing activity and schedule the real-time jobs with deadline constraints so as to subsist perpetually on a given energy source? Researchers are currently striving to design scheduling techniques which adapt to requirements of real-time energy harvesting computing systems. This paper reports very recent results that prove that the well-known Earliest Deadline First scheduler still remains an efficient option for online scheduling in these new generation computing systems. Introduction to energy harvesting systems Embedded systems can scavenge regenerative energy from human activity, ambient heat, light, radio, or vibrations [1]. So-called Energy Harvesting (EH) is a technology that allows to capture otherwise unused energy and convert it into electrical energy. Electricity can be used immediately or later through an energy reservoir such as battery or super-capacitor. The EH approach extends the life of batteries (or eliminates them entirely) and decreases maintenance. Real-Time Energy Harvesting (RTEH) systems encompass various application areas such as military, transport and medical ones. The system we target in this paper consists of a single processor unit and a rechargeable energy storage device in charge of executing a set of independent jobs with deadlines. So the problem we have to deal with is: How can we schedule the jobs so as to guarantee their timing constraints perpetually by suitably exploiting both the processor capacity and the available ambient energy.