Abstract
Much effort has been put into optimizing the energy‐efficiency of wireless \u000Asensor nodes, but existing work exclusively focuses on the transceiver and the \u000Aprocessing unit. Nevertheless, the peripheral energy consumption may dominate \u000Athat of the entire node. We introduce the concept of a dynamically scalable \u000Aperipheral voltage supply: Even for peripheral devices, a lower voltage level \u000Aleads to a lower energy consumption. Each peripheral requires a different \u000Aminimum operating voltage, but switching the voltage level consumes energy as \u000Awell. We combine theory and practice to present an algorithm weighing off the \u000Abenefits of a downscaled voltage level against the switching overhead, i. e., \u000Afor calculating an optimal peripheral voltage schedule. Our approach is capable \u000Aof self‐parameterization and has been implemented and tested on a prototype, \u000Asaving up to 47 of peripheral energy as compared to existing solutions.