Cyber Risk Prediction for UAVs in Space-Related Missions Using Deep Reinforcement Learning
Year: 2025
Type: article
Abstract: Space exploration and satellite deployment drive modern technological advancements. They are crucial for global communication, navigation, and scientific discovery. Satellites form the backbone of interstellar communication, ensuring reliable data transfer in both civilian and defense sectors. However, as space missions grow more complex, maintaining their integrity and security becomes a major challenge.Unmanned Aerial Vehicles (UAVs) play a key role in space missions. They assist in satellite deployment, orbital inspections, and inter-satellite communication. Yet, these cyber-physical systems face evolving cybersecurity threats that could jeopardize mission-critical tasks. Traditional intrusion detection systems struggle to counter the complex and dynamic cyber threats targeting UAVs in harsh space environments.This paper introduces a novel Deep Reinforcement Learning model to predict and mitigate cyber risks in space-related UAV missions. Using a publicly available dataset that combines cyber and physical UAV data, the model predicts multi-step threats such as Denial of Service, Replay, Evil Twin, and False Data Injection. This enables proactive threat mitigation. Compared to traditional machine learning models—Support Vector Machines, Random Forests, and Recurrent Neural Networks—the proposed model achieves superior performance, with 99.34% accuracy and an AUC score of 0.99.
Authors: Erfan Khosravian, Motahareh Dehghan
Institutions: Payame Noor University, Tarbiat Modares University
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7084154247
FWCI:
Cites:
Cited by:
Related to: