Clet BOUDEHENN Thesis defense

Résumé : Nowadays, more than 90% of the goods transportation is made by sea. Naval systems are strategic part of international trade and military activities. Onboard systems are increasingly computerized to optimize  operational capabilities, navigation, operations management and performance. To achieve these objectives, naval cyber systems have known a global digital transformation over the past decade. Today, information systems and operational systems have strongly converged in terms of operation. However, these major changes have made these systems vulnerable and have significantly increased the attack surface and risk of incidents. These systems have become a prime target for hackers in terms of potential harm. As a consequence, taking into account these vulnerabilities in this globalized sector must become a major issue considering the strategic, economic and geopolitical stakes. In this thesis we propose a methodology for the generation of realistic data to improve the cybersecurity of naval cybernetics systems. We have been able to develop a number of cyberattack scenarios specific to Industrial Control Systems as well as GNSS tampering cases through several case studies in response to the critical lack of data in the sector. The generation of realistic data
allowed us to propose an original feature extraction methodology adapted to anomaly detection methods in order to improve intrusion detection systems capacities. We relied on a numerical simulation platform representing all the functional and operational capacities similar to a real ship. In addition, all the experiments and tools developed during this thesis support this platform which proposes a degree of realism dedicated to training and research community.