Pulse radar system design for drone detection and deep learning using raw radar data for drone classification

Abstract

Recently, drones have become a new risk factor for society due to irresponsible flight. It is important to prevent the drones from creating dangerous situations. In this thesis, we proposed a pulse radar system for detecting drone. Since the manufactured radar system transmits fully synchronized signals, it can accurately measure the speed and distance of the object, and the performance is verified by actual drone experiment. In addition, We also proposed a deep-learning method for classifying drones using radar signals reflected on drones. We propose a new deep-learning network structure that solves the problems of existing techniques using spectrograms and verify the performance by classifying the drone using the detected drone signal