AeroTSBoost: Temporal-Statistical Boosting for Real-World UAV Telemetry Anomaly Mining

2026-05-25

AI Summary

This work addresses the challenge of anomaly detection in sparse, heterogeneous, and incomplete unmanned aerial vehicle (UAV) telemetry logs by proposing AeroTSBoost, a novel framework that integrates multiple statistical and temporal characteristics into deterministic descriptors. Specifically, after aligning multivariate time series, the method jointly incorporates distributional shifts, quantile structures, endpoint drifts, local dynamics, and lagged correlations for the first time. These enriched features are then used to train a LightGBM classifier enhanced with a class-balancing strategy. Evaluated on the UAV-SEAD and ALFA datasets, AeroTSBoost achieves AUPRC scores of 0.7516 and 0.9259, respectively, substantially outperforming existing approaches and demonstrating the effectiveness and superiority of the proposed feature fusion and modeling strategy.

Abstract

Mining anomalies from unmanned aerial vehicle (UAV) state-estimation logs is challenging because failures are sparse, temporally structured, and distributed across heterogeneous PX4 telemetry streams with variable sensor availability and missing values. We present AeroTSBoost, a temporal-statistical boosting framework for real-world UAV telemetry anomaly mining. AeroTSBoost aligns multivariate flight logs, converts each window into deterministic descriptors that capture distributional shifts, quantile structure, endpoint drift, local dynamics, and lag correlation, and trains a class-balanced LightGBM detector. On UAV-SEAD, AeroTSBoost achieves the strongest AUPRC among evaluated classical, supervised tabular, neural reconstruction, recurrent, Granger-causality-based, and frequency-domain baselines. Across five seeds, it reaches $0.7516\pm0.0043$ AUPRC and $0.5342\pm0.0108$ threshold-swept event F1, improving AUPRC by 5.79 absolute points over the strongest non-AeroTSBoost baseline. Under purged chronological and leave-log-out protocols, it remains the best AUPRC method, reaching $0.6066\pm0.0193$ and $0.6388\pm0.0315$, respectively. On related ALFA fixed-wing UAV fault logs, AeroTSBoost reaches $0.9259\pm0.0076$ leave-sequence-out AUPRC, ahead of RandomForest ($0.8835\pm0.0797$) and moments-only ($0.8700\pm0.0481$). These results show that deterministic temporal-statistical representations remain highly competitive for sparse anomaly mining in operational cyber-physical telemetry.

Problem

• UAV telemetry
• anomaly mining
• temporal structure
• sparse failures
• heterogeneous data streams

Innovation

• Temporal-Statistical Representation
• Anomaly Mining
• UAV Telemetry
• Deterministic Descriptors
• Class-Balanced Boosting