Energy-aware Reconfiguration of Sensor Nodes

Energy consumption is one of the most challenbring constraints for the design and implementation of sensor networks. As sensor nodes are commonly battery-driven, the way the available energy is used determines the lifetime of the system. In recent research, a variety of approaches has been investigated to predict energy consumption off-line using simulation, emulation, code analysis and energy models of hardware components. However, for a wide range of applications the run-time behavior is dictated by sensor measurements or external events, allowing only a worst case analysis before deployment. In this paper, we investigate different approaches to derive energy characteristics during run-time and to adapt or reconfigure the applications running on individual sensor nodes. We discuss prototype implementations based on TinyOS for the BTnode architecture and their overhead regarding energy consumption and code size.