Impact of frequency scaling on power management of an autonomous weather monitoring embedded system

Autonomous weather monitoring applications, especially the distributed ones, implies low power consumption both for sensors and computing/storage facilities. Making the autonomous system more efficient requires optimizing consumption for the computing and storage facilities are they require more energy to function than the sensors. Modern embedded systems are a solution, as they have to continuously balance the need for increased performance, increased miniaturization and decrease in power requirements for the full product to run. This is of special importance if the embedded system is designed to work outside of the normal power grid and instead rely on solar or battery power, as in the case of our autonomous weather station. Miniaturization leads to using increasingly complex system on chip devices in the embedded device in increasingly smaller nanometric geometries. These geometries favor the consumption due to leakage instead of digital transistor switching, thus minimizing the silicon clock frequency has impact in the final power consumption. This paper studies the effects of frequency in power consumption on a modern RISC based system implementation, Texas Instruments´ OMAP3 family of devices. In order to study the impact of frequency scaling alone (without the additional effects brought to by voltage scaling or methods of idling the silicon) a modern microkernel is used as the basis of a software operating system.