Time Series Prediction for Energy-Efficient Wireless Sensors: Applications to Environmental Monitoring and Video Games

Time series prediction techniques have been shown to significantly reduce the radio use and energy consumption of wireless sensor nodes performing periodic data collection tasks. In this paper, we propose an implementation of exponential smoothing, a standard time series prediction technique, for wireless sensors. We rely on a framework called Adaptive Model Selection (AMS), specifically designed for running time series prediction techniques on resource-constrained wireless sensors. We showcase our implementation with two demos, related to environmental monitoring and video games. The demos are implemented with TinyOS, a reference operating system for low-power embedded systems, and TMote Sky and TMote Invent wireless sensors.