Toward Secure Marine Navigation: A Deep Learning Framework for Radar Network Attack Detection

 Publication of new book chapter. 

Title: Toward Secure Marine Navigation: A Deep Learning Framework for Radar Network Attack Detection

📘 Book: Maritime Cybersecurity
📍 Publisher: Springer, Cham
📎 DOI: https://link.springer.com/chapter/10.1007/978-3-031-87290-7_11

Abstract:

In the evolving landscape of maritime navigation, the security of marine radar systems is paramount to ensuring safe and reliable operations. This research presents a pioneering approach to detecting and classifying network-based cyberattacks on marine radar systems through advanced deep learning techniques. Leveraging a comprehensive dataset generated from a meticulously designed simulation environment, we employ a 1D Convolutional Neural Network (1D CNN) to identify and mitigate various forms of cyber-threats. The methodology begins with the extraction of robust features from raw network traffic captured in .pcap files, encompassing a wide range of attributes. These features are crucial in characterizing normal versus malicious behaviors in radar communications. By transforming this data into a structured format suitable for machine learning, we facilitate the training of a sophisticated 1D CNN model tailored for binary and multi-class classification. The proposed model demonstrates exceptional performance, achieving near-perfect accuracy, precision, recall, and F1 scores in detecting attacks, including Denial of Service (DoS), scaling, addition, rotation, move, translation, and removal attacks. For binary classification, the model achieved perfect scores across most attack types and a near-perfect score for DoS attacks with a 99.35% F1 score. In multi-class classification, the model achieved an overall accuracy of 1.0000, with perfect precision, recall, and F1 scores for almost all classes except for a marginally lower recall (0.9900) for the DoS class. The results underscore the efficacy of deep learning in enhancing the resilience of marine radar systems against cyber-threats. This research sets a new benchmark in maritime cybersecurity by illustrating the potential of 1D CNNs in safeguarding marine radar operations.