一种卫星通信终端辐射源细微特征提取方法  

An Extraction Method for Fine Features of Satellite Communication Terminal Radiation Sources

作　　者：张海瑛[1,2] 吴玲玲 易卫明[3] 韩晓佩 李慧 ZHANG Haiying;WU Lingling;YI Weiming;HAN Xiaopei;LI Hui(The 54th Research Institute of CETC,Shijiazhuang 050081,China;Hebei Key Laboratory of Electromagnetic Spectrum Cognition and Control,Shijiazhuang 050081,China;Unit 75775,PLA,Kunming 650000,China)

机构地区：[1]中国电子科技集团公司第五十四研究所,河北石家庄050081 [2]河北省电磁频谱认知与管控重点实验室,河北石家庄050081 [3]中国人民解放军75775部队,云南昆明650000

出　　处：《无线电通信技术》2021年第3期308-314,共7页Radio Communications Technology

基　　金：国家科技计划专项经费(206Z0701G)。

摘　　要：针对卫星通信中特定网络辐射源终端的个体识别问题,提出一种基于细微特征测量和目标元数据分析相结合的特征空间构建方法。该方法充分利用卫通终端辐射源个体物理层调制的稳态特征、特定字段网络层传输的稳态特征,并结合卫星信号侦察中元数据规律和关联分析的个体特征,提升了特征维度,实现了较低信噪比条件下对细微特征参量的精确估计。同时采用深度学习方法作为分类器,经实际采集的信号数据集训练学习,达到较高的识别效果。该方法具有特征提取方法简单、识别概率高、算法稳健性好等特点,理论分析、仿真试验和工程应用均验证了该算法的有效性。A feature space construction method based on the combination of fine features measurement and target metadata analysis is proposed to solve the problem of individual identification of specific network emitter terminals in satellite communications.This method makes full use of the steady-state characteristics of individual physical layer modulation and network layer transmission in specific fields.By combining the law of metadata and the individual characteristics of correlation analysis,the dimension of feature is improved and the precise estimation of fine feature parameters is realized under the condition of low SNR.At the same time,deep learning method is used as classifier,and actual collected signal data sets are trained and learned to achieve a higher recognition effect.This method has the characteristics of simple feature extraction,high recognition probability and good robustness.Theoretical analysis,simulation experiment and engineering application have verified the effectiveness of the algorithm.

关 键 词：卫星通信终端 细微特征 特征维度 特定字段 深度学习 

分 类 号：TN911.7[电子电信—通信与信息系统]