Thesis Type: Postgraduate
Institution Of The Thesis: Middle East Technical University, Graduate School of Natural and Applied Sciences, Turkey
Approval Date: 2019
Thesis Language: English
Student: İSMAİL GÖKHAN DERE
Supervisor: Mustafa KuzuoğluAbstract:
Detection, tracking and classification of Unmanned Air Vehicles (UAVs) is an emerging and crucial capability of radars in recent years. In the presence of clutter such as a crowded city or a foggy weather above sea surface, mini and micro UAVs become very difficult for radars to detect, track and classify. Classification information of UAV targets can be very useful for the critical infrastructures in order to provide security. Examined studies imply that kinematic and characteristic features such as Doppler velocity, Radar Cross Section (RCS) fluctuations and Signal-to-Noise (SNR) are helpful features for separating drones from other moving targets. However, selection of method and extracted features have an impact on success of classification. In this study, Support Vector Machine (SVM) classification method is proposed to be used as an inclusive, useful and flexible method. Moreover, additional classification stage is proposed in this study in order to increase the success rate of separating mini/micro UAV targets from the clutter targets which has very similar characteristic properties to drones. Experiments conducted in this study for the selection of classification method and features also show that additional classification stage has an improving impact on success rate. Thus, a method is proposed in this study including comparison of classification methods, features and improvement method.