Drone Cyber Attack Detection Through Machine Learning

The use of Unmanned Aerial Vehicles (UAVs) or drones has rapidly spread across all major industries creating great instable cybersecurity challenges. With increasingly more uses such as surveillance, delivery, and environmental monitoring, UAVs operating in these applications are increasingly targeted by cyber attacks which compromise functionality, safety, and data integrity. In this research, we use machine learning (ML) techniques to detect and defend in real time cyber attacks to UAVs. We present an anomaly identification, fault data injection attacks detection and malicious activities isolation technique amongst UAV communication networks using a hybrid approach of supervised and unsupervised ML algorithms. Behavioral analysis, anomaly detection and adaptive neural network techniques are combine steps into the study to raise detection accuracy and response time. The system is shown to accurately identify each of these attacks, with little false positive, under different conditions, which are demonstrated using experimental results. The conclusion of this research emphasizes the need to incorporate machine learning for UAV cybersecurity frameworks to implement sound, real time threat recognition and mitigation.