面向空间认知通信的轻量化网络自动调制分类方法

doi:10.12305/j.issn.1001-506X.2023.07.33

摘要/Abstract

摘要：

当前自动调制分类采用的深度学习模型存在参数量与计算量大的问题。根据连续采样同相正交信号特点, 提出了一种轻量且高效的深度网络结构。通过构造方向滤波器, 首先提取相位特征, 再提取时域特征, 最后利用通道特征均值分类。采用通信信号分类数据集进行验证, 当信噪比大于0 dB时, 准确率超过60%, 信噪比大于等于10 dB时, 准确率超过90%;与主流深度模型相比, 在达到相同准确率时, 仅用20%左右的模型参数和50%左右的推理时间, 更适合被应用于空间认知通信系统。

关键词: 自动调制分类, 深度学习, 轻量化网络结构

Abstract:

The deep learning models currently used for automatic modulation classification have the problems of numerous parameters and computation. According to the characteristics of continuous sampling in-phase and quadrature signals, a lightweight and efficient deep network structure is proposed. By constructing a directional filter, the phase features are first extracted, then the temporal features are extracted, and finally the mean values of every channel features are used for classification. After validated with communication signal classification dataset, when the signal-to-noise ratio(SNR) > 0 dB, the recognition accuracy exceeds 60%, and when the SNR ≥10 dB, the recognition accuracy exceeds 90%. Compared with mainstream deep models, when reaching the same accuracy, only about 20% of the model parameters and about 50% of the actual running time are used, which is more suitable for application in space cognitive communication systems.

Key words: automatic modulation classification, deep learning, lightweight network structure

中图分类号:

崔天舒, 王栋, 黄振. 面向空间认知通信的轻量化网络自动调制分类方法[J]. 系统工程与电子技术, 2023, 45(7): 2220-2226.

Tianshu CUI, Dong WANG, Zhen HUANG. Automatic modulation classification based on lightweight network for space cognitive communication[J]. Systems Engineering and Electronics, 2023, 45(7): 2220-2226.

图/表 14

图1

图2

图3

表1

表2

图4

图5

图6

图7

图8

图9

图10

表3

表4

参考文献 31

1	BANERJEE S , SANTOS J , HEMPEL M , et al. A novel method of near-miss event detection with software defined radar in improving railyard safety[J]. Safety, 2019, 5 (3): 55. doi: 10.3390/safety5030055
2	SMITH A, EVANS M, DOWNEY J. Modulation classification of satellite communication signals using cumulants and neural networks[C]//Proc. of the Cognitive Communications for Aerospace Applications Workshop, 2017.
3	SILLS J A. Maximum-likelihood modulation classification for PSK/QAM[C]//Proc. of the Military Communications Conference, 1999.
4	WEI W , MENDEL J M . Maximum-likelihood classification for digital amplitude-phase modulations[J]. IEEE trans.on Communications, 2000, 48 (2): 189- 193. doi: 10.1109/26.823550
5	HONG L, HO K C. Identification of digital modulation types using the wavelet transform[C]//Proc. of the Military Communications Conference, 1999.
6	SWAMI A , SADLER B M . Hierarchical digital modulation classification using cumulants[J]. IEEE Trans.on Communications, 2000, 48 (3): 416- 429. doi: 10.1109/26.837045
7	HATZICHRISTOS G, FARGUES M P. A hierarchical approach to the classification of digital modulation types in multipath environments[C]// Proc. of the 35th Asilomar Conference on Signals, Systems and Computers, 2001.
8	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. Imagenet classification with deep convolutional neural networks[C]//Proc. of the 25th International Conference on Neural Information Processing Systems, 2012.
9	ARISOY E, SETHY A, RAMABHADRAN B, et al. Bidirectional recurrent neural network language models for automatic speech recognition[C]// Proc. of the IEEE International Conference on Acoustics, Speech and Signal Processing, 2015: 5421-5425.
10	ZHANG M , DIAO M , GUO L M . Convolutional neural networks for automatic cognitive radio waveform recognition[J]. IEEE Access, 2017, (5): 11074- 11082.
11	周鑫, 何晓新, 郑昌文. 基于图像深度学习的无线电信号识别[J]. 通信学报, 2019, 40 (7): 114- 125.
	ZHOU X , HE X X , ZHENG C W . Radio signal recognition based on image deep learning[J]. Journal on Communications, 2019, 40 (7): 114- 125.
12	QU Z Y , WANG W Y , HOU C B , et al. Radar signal intra-pulse modulation recognition based on convolutional denoising autoencoder and deep convolutional neural network[J]. IEEE Access, 2019, 7, 112339- 112347. doi: 10.1109/ACCESS.2019.2935247
13	PENG S L , JIANG H Y , WANG H X , et al. Modulation recognition using hierarchical deep neural networks[J]. IEEE Trans.on Neural Networks and Learning Systems, 2019, 30 (3): 718- 727. doi: 10.1109/TNNLS.2018.2850703
14	SZEGEDY C, LIU W, JIA Y Q, et al. Going Deeper with Convolutions[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2015.
15	HOCHREITER S , SCHMIDHUBER J . Long short-term memory[J]. Neural Computation, 1997, 9 (8): 1735- 1780. doi: 10.1162/neco.1997.9.8.1735
16	O'SHEA T J, CORGAN J, CLANCY T C. Convolutional radio modulation recognition networks[C]//Proc. of the International Conference on Engineering Applications of Neural Networks, 2016: 213-226.
17	WEST N E, O'SHEA T. Deep architectures for modulation recognition[C]//Proc. of the IEEE International Symposium on Dynamic Spectrum Access Networks, 2017.
18	SAINATH T N, VINYALS O, SENIOR A, et al. Convolutional, long short-term memory, fully connected deep neural networks[C]//Proc. of the IEEE International Conference on Acoustics, Speech and Signal Processing, 2015: 4580-4584.
19	O'SHEA T J , ROY T , CLANCY T C . Over-the-air deep learning based radio signal classification[J]. IEEE Journal of Selected Topics in Signal Processing, 2018, 12 (1): 168- 179. doi: 10.1109/JSTSP.2018.2797022
20	SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[EB/OL]. [2022-04-14]. https://arxiv.org/abs/1409.1556 .
21	HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778.
22	KRZYSTON J, BHATTACHARJEA R, STARK A. Complex-valued convolutions for modulation recognition using deep learning[C]//Proc. of the IEEE International Conference on Communications Workshops, 2020.
23	崔天舒, 崔凯, 黄永辉, 等. 卷积神经网络卫星信号自动调制识别算法[J]. 北京航空航天大学学报, 2022, 48 (6): 986- 994.
	CUI T S , CUI K , HUANG Y H , et al. Convolutional neural network satellite signal automatic modulation recognition algorithm[J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48 (6): 986- 994.
24	IANDOLA F N, HAN S, MOSKEWICZ M W, et al. SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and < 0.5 MB model size[EB/OL]. [2022-04-14]. https://arxiv.org/abs/1602.07360 .
25	CHOLLET F. Xception: deep learning with depthwise separable convolutions[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 1251-1258.
26	HOWARD A G, ZHU M, CHEN B, et al. Mobilenets: efficient convolutional neural networks for mobile vision applications[EB/OL]. [2022-04-14]. https://arxiv.org/abs/1704.04861 .
27	ZHANG X, ZHOU X, LIN M, et al. Shufflenet: an extremely efficient convolutional neural network for mobile devices[C]// Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 6848-6856.
28	黄知涛, 周一宇, 姜文利. 基于相对无模糊相位重构的自动脉内调制特性分析[J]. 通信学报, 2003, (4): 153- 160.
	HUANG Z T , ZHOU Y Y , JIANG W L . The automatic ana-lysis of intrapulse modulation characteristics based on the relatively non-ambiguity phase restoral[J]. Journal of China Institute of Communications, 2003, (4): 153- 160.
29	IGLESIAS V , GRAJAL J , ROYER P , et al. Real-time low-complexity automatic modulation classifier for pulsed radar signals[J]. IEEE Trans.on Aerospace and Electronic Systems, 2015, 51 (1): 108- 126. doi: 10.1109/TAES.2014.130183
30	崔天舒. 面向天基电磁信号识别的深度学习方法[D]. 北京: 中国科学院大学(中国科学院国家空间科学中心), 2021.
	CUI T S. Deep learning methods for space-based electromagnetic signal recognition[D]. Beijing: University of Chinese Academy of Sciences (National Space Science Center, Chinese Academy of Sciences), 2021.
31	O'SHEA T J, WEST N. Radio machine learning dataset ge-neration with gnu radio[C]//Proc. of the GNU Radio Confe-rence, 2016.

序号	名称	CCNet	VGG	CLDNN	ResNet
1	卷积层数	7	7	4	12
2	卷积核尺寸	(3, 1)	(8, 1)	(8, 1)	(8, 1)
3	卷积通道数	48	64	64	32
4	LSTM层数	0	0	2	0
5	LSTM单元	0	0	50	0
6	分类层层数	1	2	2	2
7	分类层单元数	48	128	128	128

数据集	RadioML
数据格式	1 024×2
SNR/dB	[-20:2:30]
调制类型数量	24
模拟调制	FM, AM-SSB-SC, AM-SSB-WC, AM-DSB-WC, AM-DSB-SC,
数字调制	OOK, OQPSK, GMSK, BPSK, QPSK, 8PSK, 16PSK, 4ASK, 8ASK, 16APSK, 32APSK, 64APSK, 128APSK, 32PSK, 16QAM, 32QAM, 64QAM, 128QAM, 256QAM
样本数据量	24×26×4 096=2 555 904

网络名称	参数数量	浮点计算量(×10⁶)	乘加操作(×10⁶)	内存使用/MB
IQCNet	43 416	25.42	12.85	2.58
IQCNet-N	43 368	50.65	25.60	4.98
CCNet	85 080	151.20	75.92	5.65
VGG	283 352	26.65	13.49	3.16
ResNet	205 848	40.38	20.39	5.65
CLDNN	168 920	-	-	-

网络结构	2080Ti	树莓派4B	Nano
IQCNet	21 330	17.83	5.86
IQCNet-N	27 490	33.55	9.58
CCNet	31 959	135.89	375.49
VGG	31 596	38.80	9.52
ResNet	28 168	47.08	18.30
CLDNN	32 486	48.44	12.89

[1]	姜雨, 袁琪, 胡志韬, 吴薇薇, 顾欣. 基于气象因素的机场进离港延误预测[J]. 系统工程与电子技术, 2023, 45(6): 1722-1731.
[2]	陈洋, 廖灿辉, 张锟, 刘建, 王鹏举. 基于自监督对比学习的信号调制识别算法[J]. 系统工程与电子技术, 2023, 45(4): 1200-1206.
[3]	张晔, 侯毅, 欧阳克威, 周石琳. 单变量序列数据分类方法综述[J]. 系统工程与电子技术, 2023, 45(2): 313-335.
[4]	邵正途, 许登荣, 徐文利, 王晗中. 基于LSTM和残差网络的雷达有源干扰识别[J]. 系统工程与电子技术, 2023, 45(2): 416-423.
[5]	李瑞泽, 张双辉, 刘永祥. 基于卷积ADMM网络的高效结构化稀疏ISAR成像方法[J]. 系统工程与电子技术, 2023, 45(1): 56-70.
[6]	韩啸, 陈世文, 陈蒙, 杨锦程. 基于互易点学习的LPI信号开集识别[J]. 系统工程与电子技术, 2022, 44(9): 2752-2759.
[7]	张立民, 谭凯文, 闫文君, 张聿远. 基于多级跳线残差网络的雷达辐射源识别[J]. 系统工程与电子技术, 2022, 44(7): 2148-2156.
[8]	金国栋, 薛远亮, 谭力宁, 许剑锟. 基于孪生神经网络的目标跟踪算法进展研究[J]. 系统工程与电子技术, 2022, 44(6): 1805-1822.
[9]	赵晓枫, 徐叶斌, 吴飞, 牛家辉, 蔡伟, 张志利. 基于全局感知机制的地面红外目标检测方法[J]. 系统工程与电子技术, 2022, 44(5): 1461-1467.
[10]	邹虹, 白陈阳, 何鹏, 崔亚平, 王汝言, 吴大鹏. 基于分布式深度学习的边缘服务放置策略[J]. 系统工程与电子技术, 2022, 44(5): 1728-1737.
[11]	陈冬, 句彦伟. 基于语义分割实现的SAR图像舰船目标检测[J]. 系统工程与电子技术, 2022, 44(4): 1195-1201.
[12]	孙晶明, 虞盛康, 孙俊. 基于深度学习的HRRP识别姿态敏感性分析[J]. 系统工程与电子技术, 2022, 44(3): 802-807.
[13]	姚云翔, 陈莹. 注意力机制下双模态交互融合的目标跟踪网络[J]. 系统工程与电子技术, 2022, 44(2): 410-419.
[14]	高涌荇, 王旭东, 汪玲, 朱岱寅, 郭军, 孟凡旺. 基于RCNN的双极化气象雷达天气信号检测[J]. 系统工程与电子技术, 2022, 44(11): 3380-3387.
[15]	周毅恒, 杨军, 夏赛强, 吕明久. 闪烁现象下旋翼目标微动参数估计方法[J]. 系统工程与电子技术, 2022, 44(1): 54-63.