当前位置: X-MOL 学术IEEE Trans. Aerosp. Electron. Sys. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
A Method for Radar Model Identification Using Time-Domain Transient Signals
IEEE Transactions on Aerospace and Electronic Systems ( IF 5.1 ) Pub Date : 2021-04-21 , DOI: 10.1109/taes.2021.3074129
Shanzeng Guo , Salman Akhtar , Anthony Mella

Radar specific emitter identification (SEI) is the process of uniquely identifying an individual emitter from the same class of radars by their individual properties that arise from hardware imperfections. However, it is challenging or perhaps impossible to generate globally unique emitter identifiers by using SEI techniques alone, due to the increasing number of radars and the subtle differences in their signal properties. We therefore introduce a multitier radar emitter identification concept that includes radar function identification, model identification, and SEI. The combination of function identifiers, model identifiers, and specific emitter identifiers could generate globally unique radar emitter identifiers. In this article, we propose to use the radio frequency (RF) features extracted from time domain transient signals for radar model identification. The RF features include the duration, maximum derivative, skewness, kurtosis, mean, variance, fractal dimension, Shannon entropy, and polynomial coefficients of the normalized energy trajectory of a transient signal, as well as the area under the trajectory curve. We propose three RF fingerprints for radar model identification, each consisting of a predetermined subset of the features. The performance of the RF fingerprints was evaluated by using five classification algorithms with two radar datasets. Our results show attractive performance with respect to hetero-model radar identification. In particular, for the hetero-model radar dataset, the pair of the all-features (AF) fingerprint and gradient boosting algorithm achieved 91.9, 90.25, and 91.1% classification accuracy for three, four, and five emitters, respectively. On this basis, we conclude that the proposed AF fingerprint could be applied directly to radar model identification by training the gradient boosting algorithm using a dataset with many radars per model.

中文翻译:


时域瞬态信号雷达模型辨识方法



雷达特定发射器识别 (SEI) 是通过硬件缺陷产生的各个属性来唯一识别同一类雷达中的单个发射器的过程。然而,由于雷达数量的不断增加及其信号特性的细微差别,仅使用 SEI 技术生成全球唯一的发射器标识符具有挑战性,甚至可能是不可能的。因此,我们引入了多层雷达发射源识别概念,包括雷达功能识别、模型识别和 SEI。功能标识符、模型标识符和特定发射器标识符的组合可以生成全球唯一的雷达发射器标识符。在本文中,我们建议使用从时域瞬态信号中提取的射频(RF)特征来进行雷达模型识别。 RF 特征包括瞬态信号归一化能量轨迹的持续时间、最大导数、偏度、峰度、均值、方差、分形维数、香农熵和多项式系数,以及轨迹曲线下的面积。我们提出了三种用于雷达模型识别的射频指纹,每个指纹都包含预定的特征子集。通过使用五种分类算法和两个雷达数据集来评估射频指纹的性能。我们的结果显示了异模雷达识别方面具有吸引力的性能。特别是,对于异模型雷达数据集,全特征 (AF) 指纹和梯度增强算法对三个、四个和五个发射器的分类精度分别达到 91.9%、90.25 和 91.1%。 在此基础上,我们得出结论,通过使用每个模型包含多个雷达的数据集训练梯度增强算法,所提出的 AF 指纹可以直接应用于雷达模型识别。
更新日期:2021-04-21
down
wechat
bug