Application of energy-harvesting in wireless sensor networks using predictive scheduling

One of today's most relevant problems in wireless sensor networks (WSN) is the supply of energy. In several cases the most promising solution could be the energy harvesting based power supply. But hard problems should be answered using that technology, because the energy available for harvesting is usually not constant and out of our control. This paper discusses the application of energy harvesting based sensors in sensor networks. A solar cell based solution is shown for sensor power supply. For optimal energy consumption the non equidistant sampling is suggested both in space and time, to decrease the unnecessary consumption. For optimal dynamic sampling the available energy, the precision needed and the prediction of the events and energy harvesting possibilities must be taken into account. Our main suggestion is that a predictive scheduling system could be used for deciding dynamically the proper timing of the sensors' measurement. This scheduling system is based on the prior knowledge of the environment and the sensors, the experimental knowledge collected already and the prediction of the future. The ideas are demonstrated analyzing the aspect of a real energy harvesting sensor design, and using a simulated motion sensor network example of a home health monitoring problem.