79. SHIMMER: AN ENERGY HARVESTING WIRELESS STRUCTURAL HEALTH MONITORING PLATFORM

Department: Computer Science & Engineering
Faculty Advisor(s): Tajana Simunic-Rosing

Primary Student
Name: Jamie Bradley Steck
Email: jlbradle@ucsd.edu
Phone: 858-534-9892
Grad Year: 2009

Abstract
Structural Health Monitoring (SHM) is the process of monitoring a structure over time and assessing the health of that structure. SHiMmer is a wireless active-sensing platform that combines localized processing with energy harvesting to provide seemingly perpetual SHM. SHiMmer monitors the local area of a structure using strategically placed piezoelectric sensors/actuators and processes the changes in the structural impedance to determine the possibility of damage. SHiMmer can eliminate the need for post-deployment physical human interaction through the use of environmental power and wireless communication with a UAV or a base station; however, the platform must adapt the accuracy, or utility, of the SHM tasks to accommodate the energy availability. For example, on sunny days, a solar-powered sensor node can perform highly accurate tasks requiring more extensive computation and communication, but on cloudy days, it must reduce utility due to a decrease in harvested energy. We develop a method to trade off energy availability, from both energy harvesting and storage, with the energy costs of the tasks the sensor node needs to accomplish. Our strategy consists of three main parts: a solar energy harvesting predictor, an energy recharge estimator, and a task controller. First, our predictor uses past energy harvesting data to predict future rates of energy harvesting. Next, our estimator determines the amount of time needed to harvest a specified amount of energy, based on actual solar power and supercapacitor testing. Finally, we present a task controller that uses two application-independent algorithms to balance task utility and execution time subject to an energy constraint. Our first algorithm determines the total execution time of a set of tasks such that desired task utilities are met. Our second algorithm solves the converse problem by approximating the maximum task utilities achievable within a global deadline. We apply our methods to a prototype Structural Health Monitoring system, demonstrating the efficiency and applicability of our algorithms and the controller's ability to adapt at runtime. Our results show that the solar energy predictor and energy recharge estimator give results within 10% of the actual values, and our controller obtains high task utility while respecting energy constraints.

« Back to Posters or Search Results