Hierarchical Intrusion Detection System for Secured Military Drone Network: A Perspicacious Approach
Vivian Ukamaka Ihekoronye, Simeon Okechukwu Ajakwe, Dong-Seong Kim, Jae Min Lee
Abstract
The significant proliferation of the Internet of drone (IoD) network due to its enormous benefits in adverse terrains has become crucial in military operations, especially for swift aerial maneuvering and combat scenarios. However, military-based IoD networks are highly susceptible to lethal attacks as a result of severed network configuration. To mitigate these security issues, recent research is focused on designing intrusion detection systems (IDS) that monitor and analyze the telemetry data that flows via the participating nodes in the IoD, hence, impeding the invasion of any malicious attack launched at the network. In this study, a hierarchical and optimized random forest (RF) anomaly-based IDS is proposed based on the randomized search cross-validation (RSCV) technique; an optimization hyper-parameter algorithm, suitable for a delay-tolerant network such as the M-IoD with consideration of the payload constraints, battery constraints and high mobility of the drones in the network. The simulation results show the superiority of the proposed model achieving the highest F1-score of 96.38%, least mean-square-error of 0.13, and adequate training time of 749ms during the prediction of the different lethal attacks when compared with other state-of-the-art machine learning classifiers subjected to similar hyperparameter tuning.