Benfriha Sihem2025-11-172025-11-172024-10-21https://dspace.univ-tlemcen.dz/handle/112/25250FANETs, or Flying Ad-Hoc Networks, are wireless communication networks comprising autonomous UAVs collaborating to fulfill various missions. FANETs are susceptible to numerous security threats. In light of this, the thesis focuses on addressing security and data privacy concerns, specifically emphasizing insider attack detection, considering drones’ unique behavior and characteristics. While numerous techniques exist to address these issues, this research delves into two main areas. First, leveraging fuzzy logic, we introduce FUBA, a robust drone behavior analytics system, to enhance trust management in FANETs. Additionally, we provide a comprehensive survey of existing techniques in this domain. Second, we propose FLID, an intelligent Intrusion Detection System (IDS) tailored for FANETs, which integrates deep learning and federated learning to detect and prevent network attacks effectively. Moreover, we enhance FLID by employing reinforcement learning for drone-client selection, thereby strengthening network security and data privacy. Our findings demonstrate that insider attack detection can be achieved without compromising data privacy, offering tangible benefits across domains such as surveillance and disaster management.enDeep learningFANETfederated learningIDSprivacyreinforcement learningsecuritytrust managementUAV. iThesis