Machine Learning-Based Real-Time Anomaly Detection for Unmanned Aerial Vehicles with a Cloud Server

Abstract

There has been an increase in the use of unmanned aerial vehicles (UAVs) for carrying out various tasks. UAVs are much smaller than manned aircrafts and do not risk the pilot's life, which makes UAVs more versatile and costeffective than regular aircrafts. However, UAVs are at risk when encountering unexpected events, such as control disruptions from intentional attacks. This raises securityand safety-related issues for the use of UAVs. To deal with such issues, this paper proposes a UAV control system that uses machine-learning technology to detect anomalies. The proposed system provides analytical redundancy by analyzing numerical data based on mathematical models. Machine learning generally requires large amounts of data to produce meaningful outcomes, and the proposed UAV control system meets these requirements since it has the ability to collect large amounts of data from UAVs. Two types of flight circumstances were simulated in an experiment. One involved the security of the UAV control system, and the other involved the safety of the UAV itself. A real-time anomaly detection system for the proposed UAV control system is also introduced and experimental results are described in detail..