Depletion-safe and efficient operation of sensor networks powered by miniature solar harvesters demands precise prediction of the future energy intake. Only recently, methods that join local knowledge and global weather forecasts have been shown to improve the prediction performance. However, making use of global weather forecasts requires network-wide dissemination with low energy and resource consumption. We present and evaluate a system architecture that automatically obtains global weather forecasts from freely available online resources and disseminates them into the sensornetwork.
We implemented our system for sensor network hardware based on TinyOS and Java. We conducted a system integration test to verify the functionality of all components and evaluated the performance of the dissemination algorithm with testbed experiments. While we find that energy consumption is negligible and resource usage is low, dissemination was reliable with a low delay compared to the hourly interval of forecast updates.