A Comparative of Detecting Physical and Cyber Attacks on Drones Using Machine Learning and Deep Learning Techniques

Abstract

Multitudes of Unmanned Aerial Vehicles (UAVs) are generally embraced in military and civilian applications. Yet, this physical-cyber method is intimidated by cyber-attacks. Recently, Machine Learning (ML) based attacks detection approaches have been effectively embraced to detect cyber-attacks. This paper presented the Intrusion Detection Security (IDS) approach. The proposed approach investigates UAVs’ cyber and physical attributes under normal process and attack circumstances. Two types of cyber-attacks have been classified: Denial-of-service (DoS) and replay. This study developed IDS approaches established on ML and Deep Learning (DL) prototypes, including Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Naive Bayes (NB), and One Dimensional Convolutional Neural Networks (1DCNN). The produced approach is trained using physical and cyber attributes individually. The finding results indicate that the 1D-CNN model achieved higher accuracy (99.79%) compared to the machine learning algorithms. The experimental results show the efficiency of the proposed method’s performance.