Time Sensitive Anti-Infoswarm Agnostic Intelligence for Safe UAV Communication
Simeon Okechukwu Ajakwe, Dong‐Seong Kim · 2024
Developing a reliable, time-sensitive, and intelligent-based security framework suitable for unmanned aerial vehicle (UAV) systems' resource-constraint uniqueness guarantees swift communication and response to scenarios. This work proposed a portable and trustworthy artificial intelligence (AI)-powered approach to address infoswarms in UAV networks. The system combines explainable AI, blockchain, and zero-trust technologies, to provide robust and resilient inter-Uavcommunication security against different facets of UAV network intrusions. The adopted lightweight explainable gradient boosting (eXGB) model was trained on a public on-the-edge dataset, to identify features of significant contribution. Also, it was compared with three (3) models in terms of reliability, efficiency, and resilience. The result shows that the eXGB model can provide resource-aware, timely, and intelligent security against different levels of intrusions in UAV swarm communication through significant feature contribution and gains of 2.34% prediction accuracy, 2.97% sensitivity, 43.1 ms latency, and 0.0013 MB storage.