A Multiagent and Machine Learning Based Denial of Service Intrusion Detection System for Drone Networks

Abstract

AbstractThe objective of this research work was to propose a model based on Muli-Agent System and on machine learning techniques to detect Denial of Service (DoS) cyber-attacks targeting the networks of drones. The proposed model is autonomous, characterized by its high performance and enables the detection of known and unknown DoS attacks in UAV networks with high accuracy and low false-positives and false-negatives rates. This approach is intended to address the security vulnerabilities of drone-based infrastructures and to show how important this topic is, given that little attention is paid by the scientific community to the security aspect of drones. The detection of DoS attacks is an indispensable security measure to ensure the high availability of drone systems typically used in emergency situations (Intelligent Health and Public Safety Systems) where geospatial information is sensitive and highly critical. The proposed approach has made it possible to detect DoS attacks using multi-agent systems and the machine learning Decision Tree algorithm, which was chosen after testing several machine learning algorithms (Such as Random Forest, Decision Tree, Tree Ensemble, Naïve Bayes, Support Vector Machine…) on the CICIDS2017 which is a reference dataset used by researchers working on Network Intrusion Detection Systems (NIDS). The results of the experiment were conclusive and demonstrated the efficiency and effectiveness of our system in detecting DoS attacks with 100% of accuracy and with null rates of false positive and false negative.KeywordsDenial of serviceGeospatial dataIntrusion detectionMachine learningNetwork of dronesCICIDS2017Muli-agent systems