面向移动多站协同雷达信号分选的时差先验增量密度聚类方法

陈金立; 樊宇; 王延杰; 张劲东

doi:10.11999/JEIT260151

面向移动多站协同雷达信号分选的时差先验增量密度聚类方法

doi: 10.11999/JEIT260151 cstr: 32379.14.JEIT260151

陈金立^1, ,,
樊宇^{1, 2},
王延杰¹,
张劲东³

1.
南京信息工程大学电子与信息工程学院南京 210044
2.
南京信息工程大学天长研究院天长 239300
3.
南京航空航天大学电子信息工程学院南京 211106

基金项目: 国家自然科学基金(62071238)，江苏省自然科学基金(BK20191399)

详细信息

作者简介:
陈金立：男，教授，研究方向为雷达辐射源信号分选、MIMO雷达信号处理，邮箱　chen820803@yeah.net

樊宇：男，硕士生，研究方向为雷达辐射源信号分选

王延杰：男，硕士，研究方向为电子侦察与雷达信号分选

张劲东：男，教授，研究方向为新体制雷达的研发、雷达信号分析与处理、高速数字信号处理系统设计与实现

通讯作者:
陈金立　chen820803@yeah.net

中图分类号: TN957.51
计量
- 文章访问数: 38
- HTML全文浏览量: 14
- PDF下载量: 4
- 被引次数: 0
出版历程
- 收稿日期: 2026-02-04
- 修回日期: 2026-07-07
- 录用日期: 2026-07-07
- 网络出版日期: 2026-07-19

An Incremental Density Clustering Method with Time-Difference Prior for Mobile Multi-Station Radar Signal Sorting

CHEN Jinli^{1
, ,},
FAN Yu^{1, 2},
WANG Yanjie¹,
ZHANG Jindong³

1.
School of Electronic and Information Engineering, Nanjing University of Information Science and Technology, Nanjing 210044, China
2.
Tianchang Research Institute, Nanjing University of Information Science andTechnology, Tianchang 239300, China
3.
College of Electronic and Information Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China

Funds: The National Natural Science Foundation of China (62071238), The National Nature Science Foundation of Jiangsu Province (BK20191399)

摘要

摘要: 在移动多站协同电子侦察场景中，观测站位置随时间变化使雷达辐射源信号的到达时间差(Time Difference of Arrival, TDOA)呈现明显的时变特性。传统多站信号分选方法通常基于固定时差模型构建，当观测站运动发生运动时，难以准确表征到达时间差的演化过程，易引发雷达簇误分裂与增批现象，导致分选稳定性下降。针对上述问题，该文提出一种面向移动多站协同雷达信号分选的时差先验增量密度聚类方法。该方法首先对多站到达时间差观测数据进行在线微簇构建，并利用微簇之间的时空关联关系实现初始聚合成宏簇；在此基础上，根据观测站运动信息建立时差随时间变化的线性先验状态模型，采用卡尔曼滤波(Kalman Filter, KF)对宏簇的时差演化轨迹进行递推预测。预测得到的时差先验被引入增量聚类更新过程，通过构建时空联合得分函数约束新到样本与微簇的关联判定。在增量聚类更新过程中，基于预测得到的时差先验，结合双重马氏距离统计判据，对发生漂移的雷达簇进行合并，从而有效抑制雷达簇增批现象。仿真结果表明，在到达时间(Time of Arrival, TOA)测量误差为150 ns、干扰率为15%的移动多站场景下，所提方法的分选正确率达到96.5%。
- 信号分选 /
- 移动多站 /
- 线性先验 /
- 卡尔曼滤波 /
- 增量密度聚类
Abstract: Objective In modern electronic warfare, radar signal sorting is a pivotal technology for electronic reconnaissance. Its primary objective is to deinterleave and categorize pulses from multiple radar emitters within dense and overlapping pulse streams. However, in practical reconnaissance missions, particularly those involving mobile platforms such as aircraft, the observation stations are in constant motion, while the target radar emitters typically remain stationary. This dynamic geometric relationship between the observation stations and the emitters causes the Time Difference of Arrival (TDOA) of intercepted signals to exhibit non-stationary characteristics that evolve over time. Conventional sorting algorithms lack the mechanism to perceive this dynamic feature drift. Consequently, continuous TDOA trajectories generated by the same emitter are prone to being incorrectly partitioned into multiple clusters during data processing, resulting in cluster proliferation and degraded sorting performance. To address these issues, an incremental density-based clustering method incorporating TDOA prior information is proposed. The proposed method effectively exploits the temporal evolution characteristics of TDOA, achieving high stability and high sorting accuracy in mobile multi-station scenarios. Methods First, by analyzing the temporal evolution characteristics of TDOA in mobile multi-station scenarios, the TDOA evolution process is approximated as a linear function of time, and a linear prior model is established to characterize its dynamic behavior. An online micro-cluster construction strategy is then adopted for multi-station TDOA samples, and macro-clusters are formed according to the spatiotemporal intersection relationships among micro-clusters. For each macro-cluster, independent Kalman filters are constructed for different TDOA dimensions. The TDOA value and its rate of change are defined as state variables, and recursive state prediction is performed to provide dynamic feature references for newly arriving samples. During sample association, a spatiotemporal joint scoring function is designed, in which prediction residuals generated by the Kalman filters are incorporated as dynamic constraints. Consequently, the matching criterion evolves from a conventional static density-based rule into a joint spatiotemporal consistency constraint, enabling more accurate assignment of newly arriving TDOA samples. To suppress cluster proliferation caused by TDOA feature evolution, a concept drift detection mechanism based on macro-cluster center evolution is further introduced. Once concept drift is detected, posterior state vectors and covariance matrices estimated by the Kalman filters are employed to construct a dual Mahalanobis-distance criterion that jointly evaluates state-distribution overlap and predicted-position overlap. Under a 95% confidence threshold, mis-split radar clusters are adaptively merged and assigned to the same radiation source, thereby effectively suppressing cluster proliferation induced by concept drift. Results and Discussions In the simulation experiments, radar pulse data distributions are generated according to the radar parameters (Table 1). The multi-station TDOA corresponding to the same radiation source exhibits a near-linear evolution trend with respect to the Time of Arrival (TOA) at the observation station(Fig. 5). To comprehensively evaluate the performance of the proposed method in complex environments, a simulation analysis on radar cluster proliferation suppression is first conducted (Fig. 6). In this scenario, radar emitters E4 and E9 are selected from the nine simulated radiation sources as representative cases for analysis. Compared with DBSCAN, Incremental Density-Based Clustering (ICDC), cloud model-based sorting, and PointNet++ sorting algorithms, the proposed method effectively suppresses radar cluster proliferation and demonstrates superior cluster stability during the sorting process. Furthermore, to further demonstrate the superiority of the proposed approach, its sorting performance was comprehensively compared with several representative algorithms, including the conventional histogram method, grid-based clustering, DBSCAN, ICDC algorithm, cloud model-based sorting, and PointNet++ sorting algorithms. The sorting accuracy of various algorithms under different TOA measurement errors (Fig. 7) indicates that the proposed method achieves a sorting accuracy exceeding 96% when the TOA measurement error ranges from 50 to 300 ns, demonstrating remarkable robustness to measurement noise. Under different pulse interference rates (Fig. 8), the sorting accuracy of the proposed method remains above 96%, exhibiting excellent interference suppression capability. To further assess the stability of different algorithms under mobile observation platforms, simulations were performed at different observation station velocities (Fig. 9). The proposed method consistently maintains a high sorting accuracy across all tested velocities. Even at relatively high observation station speeds, the sorting accuracy remains at approximately 94%, indicating that the proposed approach retains favorable stability and robustness under dynamic observation conditions. In addition, simulations on radar cluster proliferation and missed-cluster probabilities are conducted (Fig. 10), and the computational complexities of different algorithms are analyzed (Table 2). The results demonstrate that the proposed method provides a favorable balance among sorting accuracy, cluster proliferation suppression, missed-cluster control, and computational cost. Conclusions To address the sorting performance degradation caused by the temporal evolution of TDOA features in mobile multi-station scenarios, an incremental density-based clustering method incorporating TDOA prior information is proposed. By incorporating observation station motion information to construct a dynamic TDOA state model and integrating Kalman Filters to achieve recursive prediction of TDOA evolution, the proposed method effectively suppresses radar cluster proliferation and mis-splitting caused by concept drift. In addition, the spatiotemporal joint criterion and the adaptive cluster merging mechanism further enhance the robustness of the algorithm under complex and dynamic environments. Simulation results verify the effectiveness and stability of the proposed method in mobile multi-station cooperative reconnaissance scenarios. Future work will focus on real-time multi-parameter fusion-based sorting methods in complex electromagnetic environments. Furthermore, adaptive adjustment strategies for the micro-cluster spatial intersection threshold, process noise covariance matrix, and measurement noise variance will also be investigated.
- Signal sorting /
- Mobile multi-station /
- Linear prior /
- Kalman filters /
- Incremental density-based clustering(ICDC)

HTML全文

图 1 移动多站协同侦察场景图

下载: 全尺寸图片幻灯片

图 2 面向移动多站协同雷达信号分选的时差先验增量密度聚类分选框架

下载: 全尺寸图片幻灯片

图 3 宏簇形成示意图

下载: 全尺寸图片幻灯片

图 4 概念漂移引发的雷达簇增批现象示意图

下载: 全尺寸图片幻灯片

图 5 移动观测站条件下雷达辐射源TDOA随时间演化分布图

下载: 全尺寸图片幻灯片

图 6 不同算法雷达簇增批抑制效果对比图

下载: 全尺寸图片幻灯片

图 7 分选正确率随TOA测量误差的变化曲线

下载: 全尺寸图片幻灯片

图 8 分选正确率随干扰率的变化曲线

下载: 全尺寸图片幻灯片

图 9 分选正确率随观测站速度变化的柱状对比图

下载: 全尺寸图片幻灯片

图 10 不同分选算法的增批与漏批概率对比柱状图

下载: 全尺寸图片幻灯片

表 1 仿真雷达辐射源参数

辐射源序号	PRI(us)	PW(us)	CF(MHz)	BW(MHz)	发射脉冲数	位置坐标(km)
E1	450抖动	10.3固定	2500固定	7.8/7.9跳变	1296	[20，40，0]
E2	420～510滑变	10/10.6跳变	2400固定	7固定	945	[0，–70，0]
E3	480/550/620 参差	10～10.5滑变	2200/2600跳变	5.9/8.1跳变	846	[10，-30，0]
E4	1100抖动	7.8固定	2250～2400滑变	6.4～7.9滑变	543	[40，–40，0]
E5	1380～1500滑变	11固定	2400固定	8固定	426	[–25，70，0]
E6	1480固定	9.6/10跳变	2450～2550滑变	8.6固定	122	[–40，–40，0]
E7	1500/1750/2000参差	12固定	2800固定	5固定	366	[–30，10，0]
E8	8000抖动	11.7	2600/2750跳变	5～6.5滑变	75	[–40，–60，0]
E9	10000固定	7.5～8.1滑变	2300固定	7.3固定	60	[0，0，0]

下载: 导出CSV

表 2 不同算法的运行时间及分选正确率对比

分选算法	本文算法	PointNet++分选算法	云模型分选算法	ICDC聚类算法	DBSCAN聚类算法	网格聚类算法	直方图算法
运行时间(s)	8.242	6.173	9.514	6.325	26.311	8.753	8.881
分选正确率(%)	96.542	92.346	89.061	87.353	89.036	57.204	42.098

下载: 导出CSV

参考文献(19)

[1]	ZHOU Zixiang, FU Xiongjun, DONG Jian, et al. Radar signal sorting with multiple self-attention coupling mechanism based transformer network[J]. IEEE Signal Processing Letters, 2024, 31: 1765–1769. doi: 10.1109/LSP.2024.3421948.
[2]	LI Ziying, FU Xiongjun, DONG Jian, et al. Radar signal sorting via graph convolutional network and semi-supervised learning[J]. IEEE Signal Processing Letters, 2025, 32: 421–425. doi: 10.1109/LSP.2024.3519884.
[3]	ZHOU Zixiang, FU Xiongjun, DONG Jian, et al. Radar signal sorting with multiple self-attention coupling mechanism based transformer network[J]. IEEE Signal Processing Letters, 2024, 31: 1765–1769. doi: 10.1109/LSP.2024.3421948.
[4]	李松炜, 杨松岩, 邓志安, 等. 基于自适应流聚类的多站时差协同信号分选方法[J]. 信号处理, 2024, 40(4): 682–694. doi: 10.16798/j.issn.1003-0530.2024.04.007. LI Songwei, YANG Songyan, DENG Zhian, et al. Cooperative signal sorting method based on multi-station TDOA and adaptive stream clustering[J]. Journal of Signal Processing, 2024, 40(4): 682–694. doi: 10.16798/j.issn.1003-0530.2024.04.007.
[5]	马爽, 吴海斌, 柳征, 等. 基于递归扩展直方图的辐射源时差分选方法[J]. 国防科技大学学报, 2012, 34(5): 83–89. doi: 10.3969/j.issn.1001-2486.2012.05.017. MA Shuang, WU Haibin, LIU Zheng, et al. Method for emitter TDOA sorting based on recursive extended histogram[J]. Journal of National University of Defense Technology, 2012, 34(5): 83–89. doi: 10.3969/j.issn.1001-2486.2012.05.017.
[6]	刘智鑫, 赵拥军. 约束准则下扩展时差直方图脉冲分选配对方法[J]. 西安电子科技大学学报, 2019, 46(6): 102–111. doi: 10.19665/j.issn1001-2400.2019.06.015. LIU Zhixin and ZHAO Yongjun. Pulse sorting and pairing based on the constrained extended TDOA histogram[J]. Journal of Xidian University, 2019, 46(6): 102–111. doi: 10.19665/j.issn1001-2400.2019.06.015.
[7]	马贤同, 罗景青, 孟祥豪. 基于时差的同类辐射源信号分选定位方法[J]. 电子与信息学报, 2015, 37(10): 2363–2368. doi: 10.11999/JEIT141480. MA Xiantong, LUO Jingqing, and MENG Xianghao. Signal sorting and positioning method for similar radiation sources based on time difference of arrival[J]. Journal of Electronics & Information Technology, 2015, 37(10): 2363–2368. doi: 10.11999/JEIT141480.
[8]	陈涛, 王天航, 郭立民. 基于时差多参分选的多层感知器网络脉间识别[J]. 电子与信息学报, 2018, 40(7): 1567–1574. doi: 10.11999/JEIT170913. CHEN Tao, WANG Tianhang, and GUO Limin. Recognition of pulse repetition interval of multilayer percetron network based on multi-parameter TDOA sorting[J]. Journal of Electronics & Information Technology, 2018, 40(7): 1567–1574. doi: 10.11999/JEIT170913.
[9]	ZHAO Zihao, ZHANG Huaguo, and GAN Lu. A multi-station signal sorting method based on TDOA grid clustering[C]. 2021 IEEE 6th International Conference on Signal and Image Processing (ICSIP), Nanjing, China, 2021: 773–778. doi: 10.1109/ICSIP52628.2021.9689025.
[10]	XIANG Zhexi, ZHOU Zou, GU Yu, et al. A multi-station signal sorting method based on time difference pair and azimuth information[C]. Proceedings of the 12th International Conference on Computer Engineering and Networks, Singapore, 2022: 466–475. doi: 10.1007/978-981-19-6901-0_49.
[11]	LU Yang and DENG Zhian. A recursive sorting method based on multi-station TDOA vector[C]. 2023 8th International Conference on Intelligent Computing and Signal Processing (ICSP), Xi'an, China, 2023: 1024–1027. doi: 10.1109/ICSP58490.2023.10248871.
[12]	姜宏志, 赵闯, 胡德秀, 等. 基于时差多参的单脉冲信号实时配对分选[J]. 电子学报, 2021, 49(3): 566–572. doi: 10.12263/DZXB.20191316. JIANG Hongzhi, ZHAO Chuang, HU Dexiu, et al. Real-time deinterleaving algorithm for single pulse signal based on TDOAs and multi-parameter information[J]. Acta Electronica Sinica, 2021, 49(3): 566–572. doi: 10.12263/DZXB.20191316.
[13]	仇洪冰, 周兰兰, 周陬, 等. 基于云模型的辐射源信号多站分选方法[J]. 电子学报, 2022, 50(10): 2469–2477. doi: 10.12263/DZXB.20210212. QIU Hongbing, ZHOU Lanlan, ZHOU Zou, et al. Multi-station sorting method for emitter signal based on cloud model[J]. Acta Electronica Sinica, 2022, 50(10): 2469–2477. doi: 10.12263/DZXB.20210212.
[14]	陈涛, 邱宝传, 肖易寒, 等. 基于点云分割网络的雷达信号分选方法[J]. 电子与信息学报, 2024, 46(4): 1391–1398. doi: 10.11999/JEIT230622. CHEN Tao, QIU Baochuan, XIAO Yihan, et al. The radar signal deinterleaving method base on point cloud segmentation network[J]. Journal of Electronics & Information Technology, 2024, 46(4): 1391–1398. doi: 10.11999/JEIT230622.
[15]	姚璐, 孙浩禹, 李松炜, 等. 移动辐射源目标多站协同雷达信号分选[J]. 海军航空大学学报, 2025, 40(6): 932–942. doi: 10.7682/j.issn.2097-1427.2025.06.014. YAO Lu, SUN Haoyu, LI Songwei, et al. Sorting of multi-station collaborative radar signals for mobile radiation source targets[J]. Journal of Naval Aviation University, 2025, 40(6): 932–942. doi: 10.7682/j.issn.2097-1427.2025.06.014.
[16]	陆昊阳, 范玉雷, 高楠, 等. 一种适用数据流概念漂移检测与适应的增量密度聚类算法[J]. 电子学报, 2025, 53(6): 2050–2062. doi: 10.12263/DZXB.20250060. LU Haoyang, FAN Yulei, GAO Nan, et al. An incremental density-based clustering algorithm for concept drift detection and adaption over data stream[J]. Acta Electronica Sinica, 2025, 53(6): 2050–2062. doi: 10.12263/DZXB.20250060.
[17]	WANG Jiale, LIAN Jie, and ZHU Guolei. A weighted Mahalanobis distance target sensing strategy for underwater integrated sensing and communication systems[J]. IEEE Wireless Communications Letters, 2024, 13(1): 79–83. doi: 10.1109/LWC.2023.3320121.
[18]	LI Zhengbang, JIAO Yujie, FU Pan, et al. A novel method for approximating the distribution of chi-squared-type mixtures[J]. Communications in Statistics—Theory and Methods, 2025, 54(12): 3462–3475. doi: 10.1080/03610926.2024.2393703.
[19]	常安琪, 乔宏乐, 徐伟. 基于DBSCAN聚类和曲线拟合的PRI分选算法[J]. 火控雷达技术, 2023, 52(2): 21–26. doi: 10.19472/j.cnki.1008-8652.2023.02.004. CHANG Anqi, QIAO Hongle, and XU Wei. A PRI sorting algorithm based on DBSCAN clustering and curve fitting algorithm[J]. Fire Control Radar Technology, 2023, 52(2): 21–26. doi: 10.19472/j.cnki.1008-8652.2023.02.004.