An Evolutionary Energy Prediction Model for Solar Energy-Harvesting Wireless Sensor Networks

Energy harvesting plays a significance role in wireless sensor networks for it can keep the nodes surviving as long as possible, especially when the wireless sensor networks are established in somewhere that electricity is unavailable from the power station. Making use of solar energy is one solution to mitigate this problem, however, on account of the ever-changing weather conditions and the sun’s cycles, the solar energy can be very unreliable and inconstant. Thus, in this paper, a new energy prediction model named RE-prediction is presented for solar energy-harvesting wireless sensor networks, which adopts current solar energy data calculated by the ASHRAE model and the mean of last days to estimate the solar energy data in future. By comparing our RE-prediction model with other existing energy prediction models, such as EWMA, WCMA, and Pro-Energy model via the experimental analysis of these four prediction models with the same datasets, the RE-prediction model is proved to be superior to the other three in accuracy, and obtains a far smaller relative average error successfully.