LighTellite: Reinforcement Learning-Based Framework for Energy Efficient Onboard Satellite Anomaly Detection

Dataset used in the publication: "LighTellite: Reinforcement Learning-Based Framework for Energy Efficient Onboard Satellite Anomaly Detection".Please cite the paper if you use the dataset. @article{tovlightellite, title={LighTellite: Reinforcement Learning-Based Framework for Energy Efficient Onboard Satellite Anomaly Detection}, author={Tov, Aviel Ben Siman and Grolman, Edita and Elovici, Yuval and Shabtai, Asaf} } The dataset consists of telemetry and system-level measurements collected from a CubeSat testbed environment (AegisSat), designed to support research on anomaly detection and energy-aware model selection under resource constraints. Data was collected during multiple experimental runs under both nominal operation and injected attack scenarios. Each sample includes time-series features related to system performance, energy consumption, and operational state,.Attacks are labeled. The dataset is intended to support reproducibility of the results reported in the associated paper and to enable further research in satellite cybersecurity, anomaly detection, and resource-aware machine learning. The accompanying files include the raw data.While preprocessing files are provided in LighTellite's paper repository.