Operationalizing Solar Energy Predictions for Sustainable, Autonomous IoT Device Management

Abstract

For sustainable Internet of Things (IoT) systems, solar-power prediction is an essential element to optimize performance, allowing devices to schedule energy-intensive tasks in periods with excess energy. In regions with volatile weather, this requires taking the weather forecast into account. The problem is how to provide such solar energy predictions with high accuracy for large-scale IoT systems with various devices in an autonomous way, without manual adaptation effort. We present a detailed study on machine-learning approaches for the prediction of solar power intake for large scale IoT systems. We examine which machine learning models, feature sets and sampling rates gain the best results for a medium-term forecasting horizon. We also explore an operational setting in which devices are deployed without prior data and machine learning models are re-trained for each sensor continuously as a form of online learning. Our results show that prediction errors can be reduced by 20 % compared to the state of the art, despite strong weather volatility.

Modeling Periodic Energy-Harvesting Computing Systems

“…Figure 3 shows the energy-feasible region for ProtoNN in a system with a solar panel energy harvester on three representative days throughout the year. We use the solar panel power output trace by Kraemer et al [7] which is collected over a two-year period in Trondheim, Norway. Since Trondheim is close to the Arctic Circle, there is very limited sunlight in winter which result in a daily average power output of merely 0.65 mW on 3.…”

“…To validate PES for intermittent design points, we use the solar panel energy trace from Kraemer et al [7] to drive an in-house ULP-system simulator. Our error metric is absolute relative error (i.e., Error…”

Energy Harvesting Techniques for Internet of Things (IoT)

“…Solar cells offer the highest power density, of approximately 15 mW/cm2, as compared to various other energy harvesting techniques [41]. Even though solar power is uncontrollable, and the conversion efficiency is affected by the day-night cycle, seasonal changes, weather conditions, and ambient temperature, it can be predicted and modeled so that adequate strategies are adopted for assuring continuous power to electronic devices [42], [43], [44], [45], [46]. To ensure uninterrupted operation (during the periods such as night and the presence of clouds when ambient light is not available) of the device powered by solar energy harvesting components, an architecture that includes storage elements, as Figure 4 shows, is the most common [47], [48].…”

Using Long Short-Term Memory (LSTM) and Internet of Things (IoT) for Localized Surface Temperature Forecasting in an Urban Environment

“…This relationship could be useful in heat risk management to potentially reduce disruptions and delays in railway services. Kraemer et al [16] utilized an IoT system with solar power and weather forecasts to predict solar power energy. They selected relevant features from weather forecasts and trained machine learning models that generate predictions with 20% better accuracy than current state-ofthe-art predictions.…”