AEROSPACE SHANGHAI, Volume. 42, Issue 1, 186(2025)

Radar Radiation Source Classification Based on Dual-View Collaborative Clustering and Feature Spectra

Xiaodan WU*, Chaowei HUANG, Jian WANG, Hui DI, and Xiaoying GU
Author Affiliations
  • School of Electronic Information and Electrical Engineering, Shanghai Jiaotong University, Shanghai200240, China
  • show less
    Figures & Tables(20)
    Basic calculation process of the dual-view collaborative clustering and ranking methods
    Relationship between the RF and the pulse TOA of the received data
    Relationship between the pulse AOA and TOA of the received data
    Relationship between the PW and TOA of the received data
    Schematic diagram of the instantaneous characteristics of the intra-pulse signal
    Schematic diagram of the dual-spectral characteristics of the intra-pulse signa
    Reduced dimensional distribution of the pulse characteristics for different radiation sources (reduced dimension=50)
    Distribution of the original basic signal feature data
    Clustering and sorting results for the data field grids
    Sorting results of the DBSCAN
    Results of the dual-view collaborative clustering and sorting methods
    Application of the evaluation metrics system to the clustering and ranking algorithms
    True distribution of segment 6 data based on view 1
    Initial clustering distribution of segment 6 data based on view 1
    First collaborative clustering results for segment 6 data based on view 1
    Second collaborative clustering results for seg-ment 6 data based on view 1
    Third collaborative clustering results based on segment 6 data from view 1
    • Table 1. Processing steps of the dual-view collaborative clustering algorithm

      View table
      View in Article

      Table 1. Processing steps of the dual-view collaborative clustering algorithm

      输入:透视1基本信号特征数据T(1)和透视2内脉冲特征数据T(2)输出:集群标签L(视角1的数据标签L(1)和视角2的数据标签L(2))步骤:

      步骤1 对视角1的基本信号特征数据进行KPCA降维,得到子空间T'1,对视角2的脉冲内特征数据进行KPCA降维,得到子空间T'2

      步骤2 根据基于非均匀密度的有噪声应用空间聚类算法(DBSCAN)对子空间T'1T'2进行排序,并分别获得视角1和视角2的聚类标签L1L2

      步骤3 根据Jaccard系数确定L1L2之间的相似度。如果相似度大于预先设定的阈值,则聚类过程结束;否则,继续下一步。

      步骤4 根据透视1的标签L1,对透视2的数据T2应用LDA降维,得到一个新的子空间T'2,并根据透视2的标签L2,对透视1的数据T1应用LDA降维,得到一个新的子空间T'1。返回第2步。

    • Table 2. Performance comparison of different clustering and sorting algorithms

      View table
      View in Article

      Table 2. Performance comparison of different clustering and sorting algorithms

      算法ARIAMIV测量FMSSCCHI
      数据字段网格聚类0.775 10.817 80.817 90.894 60.583 090 332
      DBSCAN0.921 80.899 40.899 40.972 30.604 160 088
      双视角协同集群0.987 30.978 30.978 30.992 60.719 990 682
    • Table 3. Comparison of the dual-view and single-view clustering and ranking experiments

      View table
      View in Article

      Table 3. Comparison of the dual-view and single-view clustering and ranking experiments

      算法ARIAMIV测量FM分数SCCHI
      单视角非均匀DBSCAN0.950 30.922 10.922 10.974 60.569 268 381
      双视角非均匀DBSCAN0.987 30.978 30.978 30.992 60.719 990 682
    Tools

    Get Citation

    Copy Citation Text

    Xiaodan WU, Chaowei HUANG, Jian WANG, Hui DI, Xiaoying GU. Radar Radiation Source Classification Based on Dual-View Collaborative Clustering and Feature Spectra[J]. AEROSPACE SHANGHAI, 2025, 42(1): 186

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Speciality Discussion

    Received: Sep. 23, 2024

    Accepted: --

    Published Online: May. 13, 2025

    The Author Email:

    DOI:10.19328/j.cnki.2096-8655.2025.01.020

    Topics