A Case for Atmospheric Transmittance: Solar Energy Prediction in Wireless Sensor Nodes

In this paper, we propose four novel schemes for solar energy prediction in wireless sensor nodes. Two of the schemes (WCMA-T and ProEnergy-T)are extensions of state-of-the-art schemes, while the remaining schemes (EWMA-T and Delta-T)are new. The proposed strategies leverage the extraterrestrial solar model [5] to get better prediction accuracy compared to state-of-the-art. We restrict our scope to schemes that only employ local information. Thus, leveraging external information, such as weather forecasts, is not permitted. In our comparison, we acknowledge that wireless sensor nodes are resource constrained. In such systems, runtime computation complexity and memory footprint of the prediction schemes is of high importance. Therefore, these overheads are also considered in our comparison. ProEnergy-T achieves an average improvement of 14.5% in accuracy compared to state-of-the-art. Delta-T achieves an average improvement of 8.3% with lower runtime computation complexity.