Predictive power consumption adaptation for future generation embedded devices powered by energy harvesting sources

The number of small embedded devices is constantly growing and it is expected that there will be 50 billion of Internet connected devices in the 2020. One of the open challenges is the way of powering these devices. At the beginning they were battery powered, but now the energy is usually harvested from the environment and then accumulated. At the same time various adaptation methods have been introduced to manage the power consumption and adaptation appropriately to the energy remaining in the batteries in order to extend battery lifetime. We think that in the future, the next generation of embedded devices will take advantage of the Internet connectivity and accessibility to the weather forecasting models in order to enhance adaptability methods by analyzing the forthcoming harvested power levels. Unfortunately, these devices will use low power and lossy network protocols which may influence the quality of adaptation. In the paper, we propose the concept of the two-stage predictive power adaptation method for embedded devices that uses the weather forecast services and is designed to handle unavailability of network connection.