Energy efficient wireless sensor network protocols for monitoring and prognostics of large scale systems

In this work, energy-efficient protocols for wireless sensor networks (WSN) with applications to prognostics are investigated. Both analytical methods and verification are shown for the proposed methods via either hardware experiments or simulation. This work is presented in five papers. Energy-efficiency methods for WSN include distributed algorithms for (i) optimal routing, (ii) adaptive scheduling, (iii) adaptive transmission power and data-rate control. In the first paper a reactive optimized energy-delay sub-networking routing (OEDSR) is developed and implemented on custom motes. OEDSR provides optimal routing through consideration of available energy, delay, and distance to the destination. Results show an improvement in energy usage and extension of network lifetime. In the second paper an adaptive distributed fair scheduling protocol for multichannel networks (MC-ADFS) is developed that provides scheduling support for WSN via management of bandwidth capacity based on priority. Results for MC-ADFS demonstrate performance increases in end-to-end delay, bit-rate, and fairness. In the third paper an adaptive distributed rate and power control (ADRPC) and optimal tracking control through the generalized Hamilton-Jacobi-Bellman are developed. Lyapunov methods are used to show the stability and performance of these protocols and a comparison between these are also introduced. Finally, in the forth paper a robust observer and a prognostics scheme is developed. This robust observer is utilized as an online approximator for detecting faults in large scale systems. Additionally, a prognostics method is developed providing estimates of future values of critical physical parameters resulting in estimation of the time-to-failure (TTF) of components.