Abstract Flying ad-hoc networks (FANETs) consist of multiple unmanned aerial vehicles (UAVs) that rely on multi-hop routes for communication. These routes are particularly susceptible to grey hole attacks, necessitating swift and accurate defence to preserve the network’s quality of service. Grey hole attacks are a type of denial-of-service attack where malicious nodes selectively drop packets, disrupting the normal flow of data in the network. This paper introduces and motivates a novel dataset, FAN-GHETS24, designed for the fast classification of various grey hole attacks. The dataset is derived from sequences of packet interactions between UAVs within the network, generated through multiple simulations of FANETs. These sequences undergo post-processing via two methods: firstly, an anonymization procedure that replaces IP addresses with standard string variables, allowing for offline model training and deployment on any UAV; and secondly, the application of feature engineering techniques to format the data for machine learning model integration. The dataset’s utility is validated using a time series classification model which focuses on classifying grey hole attacks as quickly as possible.
Hutchins et al. (Fri,) studied this question.