Efficient detection of GPS Spoofing attack! BUPT team proposes a new real-time trajectory anomaly detection scheme

Civilian UAVs rely on unencrypted civilian GPS signals, making them easy prey for spoofers who broadcast slightly stronger fake satellite signals. Once the drone locks onto the counterfeit constellation it miscalculates its position and veers off mission. Existing counter-measures demand expensive multi-frequency receivers or continuous links to cellular/reference stations—requirements that are impractical for low-cost agricultural or delivery platforms.

 

Researchers from Beijing University of Posts and Telecommunications (BUPT) and Pengcheng Laboratory now report a lightweight, fully on-board solution in the Chinese Journal of Aeronautics (Volume 38, Issue 10, October 2025).  Their Motion-State-Series Trajectory Prediction and Online Anomaly Detection (MSSTP-OAD) system reframes the problem as short-term trajectory forecasting. During an offline phase, a stacked LSTM network is trained on flight logs—straight segments, turns, climbs and loiters—recorded in the open-source SITL simulator.  Each training sample is a 20-step sequence (5 s at 4 Hz) of motion-state vectors that fuse position, velocity, acceleration, attitude angles and magnetic field readings.  The network learns to predict the next five positions from the past 20 states.

 

During flight the algorithm works in two stages: 

1.  Rapid screening: every small time slot constructs a lightweight motion vector and feeds it to a first ensemble model (E₁) for a quick anomaly count. 

2.  Final decision: at the end of the detection window, a high-dimensional vector that adds LSTM-predicted positions is sent to a second ensemble model (E₂) that combines MLP, SVM and histogram-based gradient-boosting tree classifiers under a strict majority-vote rule, sharply reducing false positives.

 

Tests on 30 000 flight segments (half normal, half with 10–100 m horizontal spoofing offsets) showed: 

- Trajectory prediction R² = 0.996 (benign) / 0.994 (under attack); RMSE < 5 m even under attack. 

- Detection: accuracy 0.984, recall 0.988, F1 0.983. 

- After an alarm, a simple “return-to-waypoint” manoeuvre flew 26 % less extra distance than the baseline method.

 

The authors emphasize that current findings are simulation-based. Field campaigns employing software-defined-radio (SDR) spoofers are now under way to quantify robustness against real-world multipath, atmospheric delay, and receiver clock drift. Upcoming efforts will: 

1) fuse magnetometer and barometer data to counter potential IMU spoofing; 

2) apply quantization-aware training to shrink LSTM weights for minimal firmware overhead; and 

3) roll out a distributed variant in which neighbouring UAVs exchange ultra-light motion digests for consensus voting—boosting resilience without revealing flight plans. 

The end goal is a drop-in firmware patch for PX4 and ArduPilot that instantly retrofits existing commercial and hobby drones with a low-cost, zero-extra-hardware shield against GPS manipulation.

 

Original Source

Tianci HUANG, Huici WU, Xiaofeng TAO, Zhiqing WEI. Prediction-based trajectory anomaly detection in UAV system with GPS spoofing attack[J]. Chinese Journal of Aeronautics, 2025, 38(10): 103478. https://doi.org/10.1016/j.cja.2025.103478.

 

About Chinese Journal of Aeronautics 

Chinese Journal of Aeronautics (CJA) is an open access, peer-reviewed international journal covering all aspects of aerospace engineering, monthly published by Elsevier. The Journal reports the scientific and technological achievements and frontiers in aeronautic engineering and astronautic engineering, in both theory and practice. CJA is indexed in SCI (IF = 5.7, Q1), EI, IAA, AJ, CSA, Scopus.

---

---