Titouna Chafiq présentera ses travaux lors d’un séminaire le Vendredi 7 Mai à 10h.
Informations sur la visioconférence à venir...
Titre :Machine learning-based detection techniques of cyber-attacks in Unmanned Aerial Systems.
The use of Unmanned Aerial Vehicles (UAVs) in military and civil applications is increased in recent years. These UAVs are generally equipped with a set of sensors and follow a predefined trajectory or remotely operated by a person. The regions of deployment can be hostile environments and inaccessible such as disaster zone or military fields. Therefore, to accomplish their mission, the UAVs need to communicate with each other, with the Ground Control Station (GCS) and with the navigation satellite system. Through these infrastructures, the UAVs are prone and vulnerable to various cyber-attacks such as GPS spoofing and False Data Injection (FDI). These attacks can be launched easily by an attacker using a simple transmitter by broadcasting fake data or wrong GPS signals. These signals can mislead UAVs that receive them from their initial trajectory. In order to ensure the predefined UAV’s mission, detecting these attacks is a real challenge and permits to get a high level of flight security, high reliability, and schedule maintenance in time. To achieve this end, we proposed two separate solutions to detect these attacks using machine learning methods. The preliminary results that our proposals provide are promising in terms of a set of metrics.