We propose and evaluate an approach for the estimation of the energy consumption of sensor nodes in IoT sensing applications. The approach is based on the identification of distinct activity phases that sensor nodes repeatedly execute. The power consumption of these activity phases is measured before the nodes are deployed. The total energy consumption at runtime is then estimated by combining the measured values with timestamps captured at runtime. Therefore, the approach can take runtime adaptations of the application behavior, as necessary for adaptive sensing, into account, but without involving complex hardware measurements of power consumption at runtime. We show that the error of the estimation for selected applications is low (max. observed was 2.438%), which makes the approach very suitable for energy-aware, adaptive sensing.
展开▼