采用极化特征的通信辐射源个体识别方法  

Individual Identification Method of Communication Emitters Using Polarization Feature

作　　者：张梓轩 齐子森[1] 许华[1] 史蕴豪 梁佳[1] ZHANG Zixuan;QI Zisen;XU Hua;SHI Yunhao;LIANG Jia(Information and Navigation School,Air Force Engineering University,Xi’an 710077,China)

机构地区：[1]空军工程大学信息与导航学院,西安710077

出　　处：《西安交通大学学报》2023年第10期207-220,共14页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金重点资助项目(61971434);空军工程大学研究生创新实践基金资助项目(CXJ2022019)。

摘　　要：针对现有方法在通信辐射源空、时、频、能域特征相近甚至相同情况下,个体分类识别效果不佳的问题,提出了一种采用极化特征的通信辐射源个体识别方法。通过分析通信辐射源双极化特征表征方法,对通信辐射源的极化信号建模,根据极化信号模型,借鉴幅度-相位法思路,构建了双极化接收系统实现极化信号接收和极化特征提取;通过对极化天线振动引起的通道不一致性进行分析,建立极化通道误差模型,得到含有时变幅相误差的极化信号表示,据此设计了基于自编码器的通道时变幅相误差校正算法,在扰动数据(含时变幅相误差和噪声的数据)与原始数据(只含噪声的数据)双驱动下,学习原始数据深层特征规律,实现对扰动数据的重构,克服了极化特征对通道噪声与时变幅相误差敏感的问题;经过设置的阈值规则进行通信辐射源个体的硬判决分类。仿真实验表明,在信噪比为10 dB时,所提方法对模拟辐射源的个体识别准确率达95%以上。实测数据表明:所提方法在暗室理想情况下,对辐射源个体的识别准确率达99%以上;在实采数据上叠加高斯白噪声、模拟信噪比为10 dB时,个体识别准确率达95%以上,所提方法的有效性得到了验证。To address the problem of poor individual identification of existing methods when the spatial,temporal,frequency and energy domain features of communication emitters are similar or even the same,an individual identification method for communication emitters using polarization features was proposed.Through the analysis of the dual-polarization feature representation method of communication emitters,a polarization signal model was established.Based on the polarization signal model and the amplitude-phase method,a dual-polarization receiving system was constructed to receive polarization signals and extract polarization features.Furthermore,through the analysis of the polarization channel inconsistency caused by the vibration of the polarization antenna,a polarization channel error model was established to obtain the polarization signal representation containing time-varying amplitude-phase errors.Thus,a channel time-varying amplitude-phase error correction algorithm based on autoencoder was designed.Under the dual drive of perturbed data(data containing time-varying amplitude-phase errors and noise)and original data(data containing only noise),the deep feature patterns of the original data were learned to realize the reconstruction of the perturbed data,overcoming the sensitivity of polarization features to channel noise and time-varying amplitude-phase errors.Finally,the hard decision classification of individual communication emitters was carried out using threshold rules that have been set.Simulation experiments show that the individual recognition rate of simulated radiation sources reaches over 95%at a signal-to-noise ratio of 10 dB,and actual data shows that the proposed algorithm achieves an individual recognition rate of over 99%under ideal darkroom conditions,and a recognition rate of over 95%when Gaussian white noise is added based on real data,demonstrating its effectiveness.

关 键 词：极化 时变幅相误差 自编码器 个体识别 

分 类 号：TG156[金属学及工艺—热处理]