Lazy scheduling for energy harvesting sensor nodes

Abstract

The paper studies the case of a sensor node which is operating with the power generated by an environmental source. We present our model of an energy driven scheduling scenario that is characterized by the capacity of the node's energy storage, the deadlines and the power dissipation of the tasks to be performed. Since the execution of these tasks requires a certain amount of energy as well as time, we show that the complexity of finding useful scheduling strategies is significantly increased compared to conventional real-time scheduling. We state online scheduling algorithms that jointly account for constraints arising from both the energy and time domain. In order to demonstrate the benefits of our algorithms, we compare them by means of simulation with the classical Earliest Deadline First Algorithm. © 2006 International Federation for Information Processing.