融合深度特征的电磁频谱异常检测算法

doi:10.12263/DZXB.20210916

电子学报 ›› 2022, Vol. 50 ›› Issue (6): 1359-1369.DOI: 10.12263/DZXB.20210916

所属专题：电磁频谱智能+

融合深度特征的电磁频谱异常检测算法

彭闯, 王伦文(), 胡炜林

国防科技大学电子对抗学院，安徽合肥 230037

收稿日期:2021-07-15 修回日期:2021-12-27 出版日期:2022-06-25
- 通讯作者:
- 王伦文
- 作者简介:
- 彭闯男，1994年出生，山东曲阜人.国防科技大学电子对抗学院博士研究生.主要研究方向为深度学习、电磁环境态势认知等.E-mail: pengchuang17@nudt.edu.cn
  王伦文（通讯作者）男，1966年出生，安徽休宁人.国防科技大学电子对抗学院教授，博士生导师.主要研究方向为智能信息处理、数据挖掘、电磁环境态势认知等.
  胡炜林男，1997年出生，四川泸州人.国防科技大学电子对抗学院硕士研究生.主要研究方向为电磁态势认知、机器学习等.E-mail: hwl550@nudt.edu.cn
- 基金资助:
- 国家自然科学基金 (11975307); 国防科技创新特区项目 (19-H863-01-ZT-003-003-12)

Spectrum Anomaly Detection Algorithm Based on the Fusion of Depth Feature

PENG Chuang, WANG Lun-wen(), HU Wei-lin

College of Electronic Engineering, National University of Defense Technology, Hefei, Anhui 230037, China

Received:2021-07-15 Revised:2021-12-27 Online:2022-06-25 Published:2022-06-25
- Corresponding author:
- WANG Lun-wen
- Supported by:
- National Natural Science Foundation of China (11975307); National Defense Technology Innovation Special Zone Project (19-H863-01-ZT-003-003-12)

摘要/Abstract

摘要：

针对电磁频谱异常检测效率不高等问题，该文结合卷积神经网络（Convolutional Neural Networks，CNN）以及长短时记忆神经（Long Short-Term Memory，LSTM）网络，提出一种融合深度特征的电磁频谱异常检测算法.首先构建深度特征提取网络，该网络包含能够分级提取深度特征的两路多层CNN以及LSTM；其次通过池化、合并等操作将网络模型提取的各级深度特征进行融合，实现频谱数据预测；最后计算预测数据与真实数据的均方误差，判别频谱异常.该算法能在无监督学习的条件下，实现多种类异常状态检测的.在公开频谱数据的多个频段对算法性能进行验证，结果表明本文算法能够有效地实现电磁频谱异常检测.

关键词: 深度学习, 异常检测, 频谱预测, 特征融合

Abstract:

To solve the problem of low efficiency of electromagnetic spectrum anomaly detection, we propose a new method of spectrum anomaly detection based on depth feature fusion which combines convolutional neural networks(CNN) and long short-term memory(LSTM) Networks. Firstly, a deep feature extraction network is constructed, which includes a multi-level CNN and a LSTM. The network can extract depth features in a hierarchical manner. Then, pooling layer, concatenate layer and other operations are used to fuse the depth features to achieve high-precision prediction of spectrum data. Finally, the mean square error between the predicted data and the real data is calculated by discriminating the spectrum anomaly. The algorithm can detect multiple kinds of abnormal states under the condition of unsupervised learning. We verified our algorithm in frequency bands of public spectrum data. The results show that our algorithm can effectively realize electromagnetic spectrum anomaly detection.

Key words: deep learning, anomaly detection, spectrum prediction, fusion of depth feature

中图分类号:

TN98

彭闯, 王伦文, 胡炜林. 融合深度特征的电磁频谱异常检测算法[J]. 电子学报, 2022, 50(6): 1359-1369.

Chuang PENG, Lun-wen WANG, Wei-lin HU . Spectrum Anomaly Detection Algorithm Based on the Fusion of Depth Feature[J]. Acta Electronica Sinica, 2022, 50(6): 1359-1369.

图/表 22

图1 基于频谱预测的异常检测算法框架

图2 单个LSTM模块示意图

图3 网络模型整体结构图

图4 各频段频谱数据图

图5 频谱异常仿真实例

图6 GSM900UL频段异常检测ROC

图7 GSM900DL频段异常检测ROC

图8 GSM1800 UL频段异常检测ROC

图9 TV频段异常检测ROC

表1 各频段异常检测AUC表

模型种类	GSM900UL	GSM900DL	GSM1800UL	TV
本文算法	0.846 0	0.890 9	1	0.854 3
LSTM	0.795 2	0.852 3	0.999 6	0.835 0
ConvLSTM	0.714 3	0.745 2	0.993 0	0.787 3
RNN	0.708 7	0.777 2	0.989 2	0.850 7
VAR	0.603 3	0.735 4	0.935 3	0.712 6
CNN	0.644 1	0.698 4	0.999 7	0.834 4

表2 各模型参数量

模型种类	本文算法	LSTM	ConvLSTM	RNN	CNN
参数量	1 144 204	973 568	32 578 176	862 592	33 473

图10 本文算法GSM900DL频段异常检测ROC曲线

图11 本文算法GSM1800DL频段异常检测ROC曲线

图12 本文算法TV频段异常检测ROC曲线

表3 本文算法不同信干比下AUC值

频段	信干比/dB
频段	$- 10$	$- 5$	0	5	10
GSM900DL	0.995 9	0.966 0	0.937 0	0.890 9	0.696 6
GSM1800DL	0.971 2	0.902 2	0.819 4	0.703 6	0.635 0
TV	0.959 5	0.943 7	0.888 4	0.854 3	0.750 4

表3 本文算法不同信干比下AUC值

频段	信干比/dB
频段	$- 10$	$- 5$	0	5	10
GSM900DL	0.995 9	0.966 0	0.937 0	0.890 9	0.696 6
GSM1800DL	0.971 2	0.902 2	0.819 4	0.703 6	0.635 0
TV	0.959 5	0.943 7	0.888 4	0.854 3	0.750 4

图13 LSTM网络GSM900 DL频段异常检测ROC曲线

图14 LSTM网络GSM1800 DL频段异常检测ROC曲线

图15 LSTM网络TV频段异常检测ROC曲线

表4 LSTM网络不同信干比下AUC值

频段	信干比/dB
频段	$- 10$	$- 5$	0	5	10
GSM900DL	0.985 3	0.933 5	0.896 3	0.852 3	0.623 3
GSM1800DL	0.892 1	0.749 7	0.626 9	0.554 9	0.518 7
TV	0.954 9	0.873 6	0.854 3	0.835 0	0.710 3

表4 LSTM网络不同信干比下AUC值

频段	信干比/dB
频段	$- 10$	$- 5$	0	5	10
GSM900DL	0.985 3	0.933 5	0.896 3	0.852 3	0.623 3
GSM1800DL	0.892 1	0.749 7	0.626 9	0.554 9	0.518 7
TV	0.954 9	0.873 6	0.854 3	0.835 0	0.710 3

图16 各模型GSM900UL频段异常检测ROC曲线

图17 各模型GSM900DL频段异常检测ROC曲线

表5 各模型异常检测AUC值

频段	本文算法	卷积核为64	卷积核为16	模型不含LSTM	4层卷积模块	2层卷积模块
GSM900UL	0.846 0	0.850 2	0.712 4	0.680 1	0.847 7	0.759 1
GSM900DL	0.890 9	0.893 2	0.824 8	0.720 4	0.883 5	0.839 2

参考文献 23

1	DINGG R, JIAOY T, WANGJ L, et al. Spectrum inference in cognitive radio networks: Algorithms and applications[J]. IEEE Communications Surveys & Tutorials, 2018, 20(1): 150-182.
2	HUANGX L, WUJ, LIW F, et al. Historical spectrum sensing data mining for cognitive radio enabled vehicular ad-hoc networks[J]. IEEE Transactions on Dependable and Secure Computing, 2016, 13(1): 59-70.
3	李铭伟, 何明浩, 冷毅, 等. 一种新的电磁频谱异动检测方法[J]. 雷达科学与技术, 2020, 18(5): 479-486.
	LIM W, HEM H, LENGY, et al. A new detection method of electromagnetic spectrum anomaly[J]. Radar Science and Technology, 2020, 18(5): 479-486. (in Chinese)
4	DONGQ, YANGZ K, CHENY, et al. Exploration of singular spectrum analysis for online anomaly detection in CRNs[J]. ICST Transactions on Security and Safety, 2017, 4(12): 153516.
5	王磊, 谢树果, 苏东林, 等. 基于时间序列分析的频谱异常自主检测和稳健估计方法[J]. 电子学报, 2014, 42(6): 1055-1060.
	WANGL, XIES G, SUD L, et al. An autonomous detection and robust estimation method of spectrum anomaly based on time series analysis[J]. Acta Electronica Sinica, 2014, 42(6): 1055-1060. (in Chinese)
6	TANW A, FANGX, ZHAOL, et al. Anomaly Detection Algorithm Based on Cluster of Entropy[C]//CCF Conference on Computer Supported Cooperative Work and Social Computing. Singapore: Springer, 2019: 359-370.
7	FENGQ S, ZHANGY B, LIC, et al. Anomaly detection of spectrum in wireless communication via deep auto-encoders[J].The Journal of Supercomputing, 2017, 73(7): 3161-3178.
8	CONNM A, JOSYULAD. Radio frequency classification and anomaly detection using convolutional neural networks[C]//2019 IEEE Radar Conference(RadarConf). Boston: IEEE, 2019: 1-6.
9	RAJENDRANS, MEERTW, LENDERSV, et al. SAIFE: unsupervised wireless spectrum anomaly detection with interpretable features[C]//2018 IEEE International Symposium on Dynamic Spectrum Access Networks(DySPAN). Seoul: IEEE, 2018: 1-9.
10	LIZ J, XIAOZ J, WANGB L, et al. Scaling deep learning models for spectrum anomaly detection[C]//Mobihoc '19: Proceedings of the Twentieth ACM International Symposium on Mobile Ad Hoc Networking and Computing. New York: ACM, 2019: 291-300.
11	GUNNL, SMETP, ARBONE, et al. Anomaly detection in satellite communications systems using LSTM networks[C]//2018 Military Communications and Information Systems Conference(MilCIS). Canberra: IEEE, 2018: 1-6.
12	闻佳, 王宏君, 邓佳, 等. 基于深度学习的异常事件检测[J]. 电子学报, 2020, 48(2): 308-313.
	WENJ, WANGH J, DENGJ, et al. Abnormal event detection based on deep learning[J]. Acta Electronica Sinica, 2020, 48(2): 308-313. (in Chinese)
13	ZHOUX H, XIONGJ, ZHANGX C, et al. A radio anomaly detection algorithm based on modified generative adversarial network[J]. IEEE Wireless Communications Letters, 2021, 10(7): 1552-1556.
14	席亮, 刘涵, 樊好义, 等. 基于深度对抗学习潜在表示分布的异常检测模型[J]. 电子学报, 2021, 49(7): 1257-1265.
	XIL, LIUH, FANH Y, et al. Deep adversarial learning latent representation distribution model for anomaly detection[J]. Acta Electronica Sinica, 2021, 49(7): 1257-1265. (in Chinese)
15	RAJENDRANS, LENDERSV, MEERTW, et al. Crowdsourced wireless spectrum anomaly detection[J]. IEEE Transactions on Cognitive Communications and Networking, 2020, 6(2): 694-703.
16	RONNEBERGERO, FISCHERP, BROXT. U-Net: Convolutional Networks for Biomedical Image Segmentation[M]//Lecture Notes in Computer Science. Cham: Springer International Publishing, 2015: 234-241.
17	TANZ L, PANP S. Network fault prediction based on CNN-LSTM hybrid neural network[C]//2019 International Conference on Communications, Information System and Computer Engineering(CISCE). Haikou: IEEE, 2019: 486-490.
18	周志华. 机器学习[M]. 北京: 清华大学出版社, 2016.
	ZHOUZ H. Machine Learning[M]. Beijing: Tsinghua University Press, 2016. (in Chinese)
19	金苍宏, 董腾然, 陈天翼, 等. 融合序列分解与时空卷积的时序预测算法[J]. 电子学报, 2021, 49(2): 233-238.
	JINC H, DONGT R, CHENT Y, et al. Spatio-temporal convolutional forecasting based on time-series decomposition strategy[J]. Acta Electronica Sinica, 2021, 49(2): 233-238. (in Chinese)
20	WELLENSM. Empirical Modelling of Spectrum Use and Evaluation of Adaptive Spectrum Sensing in Dynamic Spectrum Access Networks[D]. Aachen: RWTH Aachen University, 2010.
21	YUL, CHENJ, DINGG R, et al. Spectrum prediction based on taguchi method in deep learning with long short-term memory[J]. IEEE Access, 2018, 6: 45923-45933.
22	WANY Q, CHENY Z, YANC R, et al. Similarity-based sales forecasting using improved ConvLSTM and prophet[J]. Intelligent Data Analysis, 2021, 25(2): 383-396.
23	DINGG R, WANGJ L, WUQ H, et al. On the limits of predictability in real-world radio spectrum state dynamics: From entropy theory to 5G spectrum sharing[J]. IEEE Communications Magazine, 2015, 53(7): 178-183.

[1]	李钦, 刘伟, 牛朝阳, 宝音图, 惠周勃. 低信噪比下基于分裂EfficientNet网络的雷达信号调制方式识别[J]. 电子学报, 2023, 51(3): 675-686.
[2]	范兵兵, 何庭建, 张聪炫, 陈震, 黎明. 联合遮挡约束与残差补偿的特征金字塔光流计算方法[J]. 电子学报, 2023, 51(3): 648-657.
[3]	张聿远, 闫文君, 张立民. 基于多模态特征融合网络的空时分组码识别算法[J]. 电子学报, 2023, 51(2): 489-498.
[4]	许新征, 李杉. 基于特征膨胀卷积模块的轻量化技术研究[J]. 电子学报, 2023, 51(2): 355-364.
[5]	张永梅, 孙捷. 基于动静态特征双输入神经网络的咳嗽声诊断COVID-19算法[J]. 电子学报, 2023, 51(1): 202-212.
[6]	袁海英, 成君鹏, 曾智勇, 武延瑞. Mobile_BLNet：基于Big-Little Net的轻量级卷积神经网络优化设计[J]. 电子学报, 2023, 51(1): 180-191.
[7]	王神龙, 雍宇, 吴晨睿. 基于伪孪生神经网络的低纹理工业零件6D位姿估计[J]. 电子学报, 2023, 51(1): 192-201.
[8]	李滔, 董秀成, 林宏伟. 基于深监督跨尺度注意力网络的深度图像超分辨率重建[J]. 电子学报, 2023, 51(1): 128-138.
[9]	郭晓轩, 冯其波, 冀振燕, 郑发家, 杨燕燕. 多线激光光条图像缺陷分割模型研究[J]. 电子学报, 2023, 51(1): 172-179.
[10]	贾童瑶, 卓力, 李嘉锋, 张菁. 基于深度学习的单幅图像去雾研究进展[J]. 电子学报, 2023, 51(1): 231-245.
[11]	何滢婕, 刘月峰, 边浩东, 郭威, 张小燕. 基于Informer的电池荷电状态估算及其稀疏优化方法[J]. 电子学报, 2023, 51(1): 50-56.
[12]	王硕, 王坚, 王亚男, 宋亚飞. 一种基于特征融合的恶意代码快速检测方法[J]. 电子学报, 2023, 51(1): 57-66.
[13]	吴靖, 叶晓晶, 黄峰, 陈丽琼, 王志锋, 刘文犀. 基于深度学习的单帧图像超分辨率重建综述[J]. 电子学报, 2022, 50(9): 2265-2294.
[14]	李雪莹, 王田路, 梁鹏, 王翀. 基于系统模型的用户评论中非功能需求的自动分类[J]. 电子学报, 2022, 50(9): 2079-2089.
[15]	琚长瑞, 秦晓燕, 袁广林, 李豪, 朱虹. 尺度敏感损失与特征融合的快速小目标检测方法[J]. 电子学报, 2022, 50(9): 2119-2126.

融合深度特征的电磁频谱异常检测算法

Spectrum Anomaly Detection Algorithm Based on the Fusion of Depth Feature

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 22

参考文献 23

相关文章 15

编辑推荐

Metrics