On the Impact of an Antenna Field of View on the Classification of UAVs

Detection and classification of Unmanned Air Vehicles (UAVs) at a distance have become important because of the potential threats of the illegal usage of them. Radar systems are preferred for UAVs detection because of their advantages over other UAVs detection systems. In this paper, an investigation of the effect of an antenna Field of View (FoV) on Machine Learning (ML) accuracy is conducted. A full-wave Electromagnetic (EM) CAD tool is used to generate the required datasets for this investigation. Five UAVs were used in this work, a fixed-wing, a helicopter, two quadcopters, and a hexacopter UAVs. The ML algorithm was trained on a relative angle of 0 degrees between the UAVs and the antenna, and it was tested on relative angles of 20 degrees, 40 degrees, 60 degrees, 80 degrees, and 90 degrees between the UAVs and the antenna. The ML classification accuracy decreases with the increase of the relative angle between the UAVs and the antenna. The accuracy of a classifier can be estimated by employing Multiple-input Multiple-output (MIMO) radars to detect the Angle of Arrival (AoA) of drones and the relative angle between the drones and the antenna.