Statistical models for harvested power from human motion

We investigate the statistical properties of harvested power from a human motion based harvesting device. Based on empirical measurements, during normal daily activities, we provide models for the probability density function and the autocorrelation coefficient of the harvested power. The empirical measurements are collected from an accelerometer attached to the subject´s ankle. The measured acceleration is converted to the harvested power assuming a velocity-damped resonant generator harvester architecture before statistical modeling.