Hierarchical Ensemble Pipeline Detects Anomalies in ESA Satellite Telemetry

A new research paper introduces a hierarchical ensemble pipeline designed to detect anomalies in multivariate telemetry data from the European Space Agency (ESA) ¹. The pipeline integrates shapelet-based and statistical feature extraction, per-channel modeling, intra-channel stacking, and a final cross-channel aggregation ¹.

The research, published on arXiv, focuses on addressing the challenge of identifying unusual patterns within the complex data streams generated by satellites ¹.

The pipeline was trained and validated using time-series cross-validation and two-level masking strategies to prevent information leakage, ensuring the model's reliability ¹.

The study's authors, Lorenzo Riccardo Allegrini and Geremia Pompei, highlight the effectiveness of their hierarchical modeling approach in detecting subtle anomalies within realistic satellite telemetry data ¹.

The research paper, titled "A Hierarchical Ensemble Pipeline for Anomaly Detection in ESA Satellite Telemetry," was submitted on April 22, 2026 ¹.

The paper's abstract details the pipeline's structure, emphasizing its ability to process and analyze the complex telemetry data provided by ESA ¹.

The pipeline's performance was evaluated using the European Space Agency Anomaly Detection Benchmark (ESA-ADB) challenge ¹.

The results demonstrated strong generalization capabilities, showcasing the pipeline's ability to accurately identify anomalies ¹.

The research achieved second place in the final round of the Spacecraft Anomaly Challenge on the ESA dataset ¹.

The authors also noted that their approach ranked first on the Kaggle public leaderboard and third on the private leaderboard ¹.

The paper is categorized under Machine Learning (cs.LG) and Computer Vision and Pattern Recognition (cs.CV) ¹.

The paper is published in Communications in Computer and Information Science 2842 (2026) Chapter 7 ¹.

The research contributes to the growing field of machine learning applications in space-related data analysis, offering a robust method for anomaly detection in critical satellite telemetry ¹.