AI-Driven Cybersecurity for Satellite Systems: A Survey of Threats, Applications, Challenges and Future Directions

Abstract

Satellite systems have become increasingly important for communication, navigation, defence, emergency management, Earth observation, finance, and other critical infrastructure services. These systems are increasingly connected to ground stations, cloud platforms, Commercial Off-the-Shelf (COTS) systems, open-source software, and Artificial Intelligence (AI) and autonomous systems, thereby increasing their cyberattack surface. This survey explores the threats, applications, challenges, and future directions of AI in satellite-based cybersecurity in this survey. It highlights two recent studies, one of which combines Quantum Key Distribution (QKD) with CatBoost Machine Learning (ML) for secure satellite communications, and the other of which is a taxonomy of cyber vulnerabilities, spanning legacy, operational, and AI-enabled space systems. The review concludes that AI can be used for anomaly detection, threat classification, risk prioritisation and decision support, and quantum cryptography can enhance key exchange. To ensure future progress, there is a need for explainable AI and resilient security frameworks.