计算机科学
雷达
信号子空间
聚类分析
雷达信号处理
子空间拓扑
信号处理
人工智能
模式识别(心理学)
算法
电信
噪音(视频)
图像(数学)
作者
Qiang Guo,Shuai Huang,Li-Chao Liu,Микола Михайлович Калюжний,Stepan Douplii
标识
DOI:10.1109/taes.2024.3481212
摘要
In the modern information battlefield, radar signal deinterleaving plays a pivotal role in perceiving the electromagnetic spatial situation and grasping the initiative of the battle. The increasingly complex electromagnetic environment of modern battlefields leads to serious feature distribution space overlap of radar signal parameters, resulting in limited intraclass aggregation and interclass separability. Therefore, this article introduces the concept of a sparse subspace and proposes a radar signal deinterleaving method based on enhanced sparse subspace clustering. The generalized orthogonal matching pursuit algorithm is used in this study to construct a complex network on the basis of the self-expressiveness property among radar pulses in the same subspace, reducing the connection of pulses from different radars and thus decreasing computational complexity. The graph segmentation concept is incorporated into this algorithm, and Laplacian spectrum decomposition is employed to mine the geometric correlation from radar pulse signals. This approach enables efficient radar signal deinterleaving under nonideal conditions. Simultaneously, an algorithm based on an energy-corrected splitting threshold is proposed to calculate the number of cluster centers accurately. The proposed method effectively addresses the deinterleaving error issue caused by overlapping or missing working state parameters in intercepted complex radar pulse signals, achieving a deinterleaving accuracy of 94.74% for ten interleaved radar pulse signals with severe parameter feature overlap.
科研通智能强力驱动
Strongly Powered by AbleSci AI