Physics-Aware Processing of Rotational Micro-Doppler Signatures for DBN-Based UAS Classification Radar
Physics-Aware Processing of Rotational Micro-Doppler Signatures for DBN-Based UAS Classification Radar
复制标题
DOI：
10.1109/rfid49298.2020.9244873
复制
发表时间：
2020-09
期刊：
影响因子：
--
通讯作者：
A. Madanayake;G. Mendis;V. Ariyarathna;S. Pulipati;Tharindu Randeny;S. Bhardwaj;Xin Wang;S. Mandal;Jin Wei
中科院分区：
文献类型：
--
作者：
A. Madanayake;G. Mendis;V. Ariyarathna;S. Pulipati;Tharindu Randeny;S. Bhardwaj;Xin Wang;S. Mandal;Jin Wei
本文介绍了基于多普勒雷达的微型无人机(UAS)精确、稳健检测的硬件、信号处理和机器学习方法。在2.4 GHz连续波(CW)雷达和各种商用微型无人机设备的空中测试中，∼的典型探测准确率达到了98%。描述了几种进一步提高检测性能的方法，包括与均匀圆形阵列进行多波束合成以提供360°方位灵敏度；在毫米波频率(28 GHz)使用介质透镜天线和焦平面阵列以提高空间分辨率；以及基于多光谱的特征提取方法以改进对测量的多普勒信号中的非线性相位调制过程的建模。
This paper describes hardware, signal processing, and machine learning methods for Doppler radar-based accurate and robust detection of micro unmanned aerial systems (UAS). Typical detection accuracy of ∼98% was obtained in over-the-air tests with a 2.4 GHz continuous wave (CW) radar and a variety of commercially-available micro-UAS devices. Several methods are described for further improving detection performance, including multi-beam synthesis with uniform circular arrays to provide 360° azimuthal sensitivity; dielectric lens antennas and focal plane arrays at mm-wave frequencies (28 GHz) for improved spatial resolution; and polyspectra-based feature extraction methods for improved modeling of nonlinear phase modulation processes within the measured Doppler signatures.