电信科学 ›› 2022, Vol. 38 ›› Issue (1): 25-35.doi: 10.11959/j.issn.1000-0801.2022011

• 研究与开发 • 上一篇    下一篇

基于频谱形状的低复杂度雷达信号分类

尹良1, 林睿1, 王晓雷2, 姚宇亮1, 周林1, 何元1   

  1. 1 北京邮电大学信息与通信工程学院,北京 100876
    2 军事科学院国防科技创新研究院,北京 100071
  • 修回日期:2022-01-09 出版日期:2022-01-20 发布日期:2022-01-01
  • 作者简介:尹良(1983- ),男,博士,北京邮电大学副教授、硕士生导师,主要研究方向为信号检测与估计、软件无线电、机器学习在信号识别中的应用、无线电频谱工程半实物仿真
    林睿(1998- ),男,北京邮电大学信息与通信工程学院硕士生,主要研究方向为雷达信息处理、人工智能
    王晓雷(1982- ),男,博士,军事科学院国防科技创新研究院副研究员、硕士生导师,主要研究方向为智能信号处理、认知电磁对抗、网络通信安全
    姚宇亮(1997- ),男,北京邮电大学信息与通信工程学院硕士生,主要研究方向为电磁兼容测试、智能干扰分析、人工智能
    周林(1998- ),男,北京邮电大学信息与通信工程学院硕士生,主要研究方向为雷达信息处理、人工智能
    何元(1984- ),男,博士,北京邮电大学副教授、硕士生导师,主要研究方向为电子侦察对抗、雷达信息处理、人工智能
  • 基金资助:
    国家自然科学基金青年基金资助项目(61801034);国家重点研发计划项目(2018YFB1800802)

Low complexity radar signal classification based on spectrum shape

Liang YIN1, Rui LIN1, Xiaolei WANG2, Yuliang YAO1, Lin ZHOU1, Yuan HE1   

  1. 1 School of Communication and Information Engineering, Beijing University of Posts and Telecommunications, Beijing 100876, China
    2 National Innovation Institute of Defense Technology, Academy of Military Science, Beijing 100071, China
  • Revised:2022-01-09 Online:2022-01-20 Published:2022-01-01
  • Supported by:
    The National Nature Science Foundation Youth Fund of China(61801034);The National Key Research and Development Program of China(2018YFB1800802)

摘要:

摘 要:为解决雷达信号调制识别中存在的计算复杂度高、低信噪比环境识别准确率低和仿真数据真实度低等问题,提出了基于频谱形状的低复杂度雷达信号分类算法。对信号频谱进行归一化,按频谱采样的方法提取特征参数,训练机器学习分类模型。雷达信号源生成数据的测试结果表明,本算法对Barker码、Frank码、LFM、BPSK、QPSK 调制和常规雷达信号的分类准确率大于 90%(SNR≥3 dB),计算复杂度低,能适应信号参数变化,具有很好的泛化性。

关键词: 频谱形状, 低复杂度, 特征提取, 频谱采样

Abstract:

In order to solve the problems of high computational complexity, low recognition accuracy of low signal to noise ratio (SNR) environment and low fidelity of simulation data in radar signal modulation recognition, a low complexity radar signal classification algorithm based on spectrum shape was proposed.Signal spectrum was normalized, feature parameters were extracted by spectrum sampling method, and then machine learning classification model was trained.The test results of the data generated by the radar signal source show that the classification accuracy of Barker code, Frank code, LFM code, BPSK, QPSK modulation and conventional radar signals is more than 90% (SNR≥3 dB).The algorithm has low computational complexity, can adapt to the change of signal parameters, and has good generalization.

Key words: spectrum shape, low complexity, feature extraction, spectrum sampling

中图分类号: 

No Suggested Reading articles found!