Statistical Models for Harvested Power From Human Motion

This paper investigates the statistical properties of human motion-based harvested power, and provides models for the distribution, auto-correlation and cross-correlation of harvested power at different body locations, namely left wrist, right wrist, left ankle and waist. The models are developed based on empirical acceleration measurements while the subjects perform unscripted daily tasks. The measured accelerations are converted to harvestable power by assuming a velocity-damped resonant harvesting generator. The provided models enable realistic analysis and simulation of wearable communication systems with motion-based energy harvesting.