AI Opportunities for Increased Energy Autonomy of Low Power IoT Devices
Dimitri Galayko, Armine Karami, Philippe Basset, Elena Blokhina · 2019
This paper is a focus paper opening the special session “Ultra-low power pattern recognition for smart IoT applications”. The goal of this paper is to provide a review of the trends in the use of the recently available deep learning techniques for the leveraging of the energy autonomy of low power wireless devices used in Internet-of-Things networks. The paper discusses two families of applications of Artificial Intelligence in power optimisation: (i) Context-aware adaptive optimisation of the consumed power by IoT devices and networks by forecasting the future of the power consumption and the future of the energy available through the battery/energy harvesting, and (ii) Optimisation of energy harvesting devices in the context where available environment energy is characterized by stochastic and/or irregular patterns. The second case is illustrated by the presentation of a novel concept of kinetic energy harvesting generating electricity out of vibrations related to human body motion.