| Abstract | GPS spoofing poses a critical threat to maritime autonomous surface ships, compromising navigation integrity and situational awareness. Existing research is limited by the lack of realistic datasets and reproducible evaluation environments. In this work, we present a comprehensive framework for GPS spoofing detection, combining a modular simulation of spoofing attacks(ghost vessel, gradual drift, location jumps, replay and meaconing), machine learning-based detection, and an automated response module. Our simulator generates over 950 labeled spoofed AIS points merged with normal data to create a ground-truth dataset suitable for model training and evaluation. Among the evaluated models, a GRU-based approach achieved the best performance, with an F1-score of 0.98, high recall, and only six false negatives. The integrated response module applies debouncing logic to classify suspicious events and triggers email alerts for confirmed spoofing, enabling real-time operational monitoring. These results demonstrate that our framework provides a scalable and effective reproducible solution for enhancing maritime navigation security. |
|---|
