Tasmanian devil optimization with deep autoencoder for intrusion detection in IoT assisted unmanned aerial vehicle networks

Abstract

Background: Recently, a developing count of physical objects is linked to the Internet at an unprecedented rate, calcifying the knowledge of the Internet of Things (IoT). In several paradigms of IoT applications, unmanned aerial vehicles (UAVs) and satellites for IoT have concerned much attention and are experiencing quick progress. As for UAVs, because of their superiority in maneuverability and cost, it is established an increasingly extensive consumption in several IoT scenarios like disaster relief, rapid transportation, and environment monitoring. Security remains a main problem in the IoT supported UAV networks that are solved by the employ of intrusion detection system (IDS) methods. Objective: This article aims to present a Tasmanian Devil Optimization with Deep Autoencoder for Intrusion Detection System (TDODAE-IDS) technique in IoT assisted Unmanned Aerial Vehicle Networks. Methods: The presented TDODAE-IDS technique majorly concentrates on the effectual identification of the intrusions in the IoT based UAV networks. To accomplish this, the presented TDODAE-IDS system designs a new TDO algorithm for the feature subset selection process. Moreover, the DAE model classifies the existence of intrusion in the UAV network and the hyperparameter tuning of the DAE model takes place using the dragonfly algorithm (DFA). Results: The simulation results of the TDODAE-IDS approach were tested on a benchmark IDS dataset and the results are assessed under several measures. Conclusion: The comprehensive comparative analysis highlighted the enhanced outcomes of the TDODAE-IDS algorithm over other recent approaches with maximum accuracy of 99.36%. Therefore, the proposed model can be employed to accomplish security in the IoT assisted UAV networks.

Keywords

• Internet of things
• Unmanned aerial vehicles
• Deep autoencoder
• Intrusion detection
• Feature selection