드론과 조류의 분별을 위한 효과적인 미세도플러 특성벡터 연구

Abstract

Conventional drones have been used for a few specific purposes with technical limitations and high prices. In recent years, however, the aforementioned problems have been alleviated to a large extent, making it easier for individuals to use them and for various purposes. However, the drones can go where people can not approach without difficulty, so the possession of the abuse is increasing. In particular, existing security systems do not target the public to humans, so the need for technology against the drones is particularly prominent. However, drones are difficult to distinguish because they are similar in size and speed to tides and the Radar Cross Section (RCS) is very small. Therefore, it is necessary to identify each other using fine Doppler due to the proper motion of birds and drone. Micro-Doppler uses the Doppler effect in which the frequency of the radar signal is modulated and reflected at the relative speed of the target to the radar. The velocity of the target portion in the direction of the line of sight (LOS) of the radar using a short-time Fourier transform (STFT) Can be represented as a time-frequency 2D image. The image implies an instantaneous change in motion of the target portion over time and can discover unique features such as the motion cycle of the target. In this paper, we propose extracting Micro-Doppler from micro-motion of drones and birds to obtain several feature for discrimination. And the efficiency of the proposed method is verified through simulations using Physical Optics (PO) based modeling which is similar to realistic point scattering modeling. The reflection signals of the two objects being formed were generated by randomly selecting each variable within a certain range. In order to reduce the measurement error, 300 repetitive simulations were performed in each SNR environment. As a result, mono-static with SNR> 15dB showed more than 90% discrimination ability in two features and bi-static more than 90% in one characteristic vector. In addition, another feature showed more than 97% discrimination ability over 15dB and specific elevation interval.