Energy aware data management on AVR micro controller based systems

Abstract

Data management systems comprise various algorithms for efficiently retrieving and managing data. Typically, algorithm efficiency or performance is correlated with execution speed. However, the uptime of battery-powered mobile- and embedded systems strongly depends on the energy consumption of the involved components. This paper reports our results concerning the energy consumption of different implementations of sorting and join algorithms. We demonstrate that high performance algorithms often require more energy than slower ones. Furthermore, we show that dynamically exchanging algorithms at runtime results in a better throughput if energy is limited.