Perpetual Data Collection with Energy-Harvesting Sensor Networks

Abstract

A sustainable, uniform, and utility-maximizing operation of energy-harvesting sensor networks requires methods for aligning consumption with harvest. This article presents a lightweight algorithm for online load adaptation of energy-harvesting sensor nodes using supercapacitors as energy buffers. The algorithm capitalizes on the elementary relationship between state of charge and voltage that is characteristic for supercapacitors. It is particularly designed to handle the nonlinear system model, and it is lightweight enough to run on low-power sensor node hardware. We define two energy policies, evaluate their performance using real-world solar-harvesting traces, and analyze the influence of the supercapacitor’s capacity and imprecisions in harvest forecasts. To show the practical merit of our algorithm, we devise a load adaptation scheme for multihop data collection sensor networks and run a 4-week field test. The results show that (i) choosing a duty cycle a priori is infeasible, (ii) our algorithm increases the achievable work load of a node when using forecasts, (iii) uniform and steady operation is achieved, and (iv) depletion can be prevented in most cases.