Publication Type : Journal Article
Publisher : Institute of Electrical and Electronics Engineers (IEEE)
Source : IEEE Access
Url : https://doi.org/10.1109/access.2026.3699365
Campus : Bengaluru
School : School of Engineering
Department : Electronics and Communication
Year : 2026
Abstract : Satellites operate in a remote and inaccessible environment, making in-orbit maintenance and repair extremely challenging. Continuous telemetry is therefore monitored to assess spacecraft health and support timely decision-making. With the availability of large volumes of historical telemetry data, data-driven approaches have become essential for automated health monitoring and early anomaly warning. This paper presents a forecasting-based early anomaly detection framework for satellite telemetry. The proposed approach first predicts future telemetry values and then detects anomalies based on violations of statistically derived upper and lower control limits. Classical time-series models, namely Auto Regressive Integrated Moving Average (ARIMA) and Seasonal Auto Regressive Integrated Moving Average (SARIMA), are evaluated alongside deep learning models, including Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks. A case study using offline telemetry from the Dynamically Tuned Gyro (DTG) subsystem of a decommissioned satellite is presented. The models are trained on historical data to forecast upcoming telemetry values, and the predicted values are used to provide early warning of anomalies. The results demonstrate that the proposed forecasting-based detection approach can successfully provide advance indication of anomaly occurrence, with predicted anomaly timing closely matching the actual event. The case study focuses on a known anomaly observed in the DTG telemetry parameter during satellite operation, with the anomaly occurrence time verified using mission operation logs, operator annotations, and official anomaly reports. The proposed model successfully generated an anomaly alert, demonstrating a one-day early warning capability. The presented framework can be extended to real-time monitoring of critical telemetry parameters for in-service satellites, enabling proactive health management and improved operational reliability.
Cite this Research Publication : Puli Shravan Kumar, T. K. Ramesh, J. Krishna Kishore, Aravinda Koithyar, Satellite Anomaly Prediction Using Time-Series Analysis of Telemetry Data, IEEE Access, Institute of Electrical and Electronics Engineers (IEEE), 2026, https://doi.org/10.1109/access.2026.3699365