Energy and task management in energy harvesting wireless sensor networks for structural health monitoring

Energy harvesting sensor nodes reduce the need for post-deployment physical human interaction by using environmental power and wireless communication; however, they must adapt performance to accommodate the energy availability. This thesis presents three application-independent algorithms that adapt performance based on energy availability for steady and external trigger state conditions. Steady state operation describes the periodic execution of a set of tasks on the system. For steady state operation, a method is presented that adapts the execution rate to achieve high performance while maintaining sufficient energy. External trigger state operation occurs when an external device makes a request to the system. For external trigger state operation, algorithms are used to determine the execution time, energy consumption and performance of the request. These methods are applied to SHiMmer, a wireless, energy-harvesting structural health monitoring platform. Unlike other sensor systems that periodically monitor a structure and route information to a base station, SHiMmer is designed to acquire data using active sensing and process it locally before communicating with an external device. Results from this application demonstrate the controller's ability to adapt at runtime and maintain sufficient energy. Steady state results show that the execution rate changes with weather conditions. On average, the execution rate on a sunny day increases by 62% compared to the rate on cloudy days. External trigger state results show that processing significantly affects the efficiency of a structural health monitoring system; specifically, complex processing requires 17 times less execution time and 2.5 times less energy than transmitting raw data.