低信噪比下基于深度学习TCNN-MobileNet的调制识别

doi:10.19678/j.issn.1000-3428.0068243

摘要/Abstract

摘要：

将深度学习算法应用于调制识别任务是近年来通信领域的一个研究热点, 但现有方法存在网络复杂度高、硬件要求高、在低信噪比(SNR)下识别准确率不高等问题。结合离散小波变换方法, 提出一种基于双路卷积神经网络级联可分离卷积网络(TCNN-MobileNet)的调制识别方法。首先通过小波变换对数据进行预处理, 将信号作为输入传送到双路卷积神经网络中进行不同维度的特征提取; 然后通过融合层进行特征融合并送入轻量级神经网络MobileNetV1中, 进行调制识别模型训练; 最后通过全连接层进行11种调制识别的分类输出。在公开数据集RML2016.10a上的实验结果表明, 在-20 dB的低SNR下TCNN-MobileNet的识别准确率可达88.71%, 在18 dB的高SNR下识别准确率可达96.66%, SNR在-20~18 dB范围内时平均识别准确率为88.37%, 相比于ResNet18、ResNet34等经典网络架构提升了约35%。TCNN-MobileNet识别方法在保证识别精度不变的情况下能够降低训练参数量以及网络训练时间, 有效简化网络架构, 降低对硬件设备的要求, 对轻量级神经网络在调制识别中的应用具有借鉴意义。

关键词: 调制识别, 卷积神经网络, 小波变换, 深度学习, 低信噪比

Abstract:

In recent years, the application of deep learning algorithms to modulation recognition tasks has become a research hotspot in the field of communication. However, existing methods suffer from high network complexity, high hardware requirements, and low recognition accuracy under low Signal-to-Noise Ratio (SNR). A modulation recognition method based on a two-way convolutional neural network cascade separable convolutional network called TCNN-MobileNet is proposed, which combines the Discrete Wavelet Transform (DWT) method. First, the data are preprocessed using Wavelet Transform (WT), and the signal is transmitted as input to a dual convolutional neural network for feature extraction in different dimensions. Subsequently, feature fusion is performed through the fusion layer and fed into the lightweight neural network MobileNetV1 for modulation recognition model training. Finally, 11 types of modulation recognition are classified and output through a fully connected layer. The experimental results on the publicly available dataset RML2016.10a show that the recognition accuracy of TCNN-MobileNet can reach 88.71% at a low SNR of -20 dB, 96.66% at a high SNR of 18 dB, and an average recognition accuracy of 88.37% in the SNR range of -20 dB to 18 dB, which is approximately 35% higher than classical network architectures such as ResNet18 and ResNet34. The TCNN-MobileNet recognition method can reduce the number of training parameters and the network training time while ensuring that the recognition accuracy remains unchanged, effectively simplifying the network architecture and reducing hardware requirements. This is significant for the application of lightweight neural networks in modulation recognition.

Key words: modulation recognition, Convolutional Neural Network(CNN), Wavelet Transform(WT), deep learning, low Signal-to-Noise Ratio(SNR)

牛瑞婷, 严天峰, 高锐, 王映植. 低信噪比下基于深度学习TCNN-MobileNet的调制识别[J]. 计算机工程, 2024, 50(7): 204-215.

Ruiting NUI, Tianfeng YAN, Rui GAO, Yingzhi WANG. Deep Learning TCNN-MobileNet-Based Modulation Recognition Under Low Signal-to-Noise Radio[J]. Computer Engineering, 2024, 50(7): 204-215.

https://www.ecice06.com/CN/Y2024/V50/I7/204

图/表 17

图1 常见小波基的去噪效果

Fig.1 Denoising effect of common wavelet bases

图2 Coiflets小波基去噪效果对比

Fig.2 Comparison of denoising effect of Coiflets wavelet bases

图3 离散小波变换

Fig.3 Discrete wavelet transform

图4 传统卷积示意图

Fig.4 Schematic diagram of traditional convolution

图5 深度卷积示意图

Fig.5 Schematic diagram of deep convolution

图6 逐点卷积示意图

Fig.6 Schematic diagram of pointwise convolution

图7 TCNN-MobileNet网络架构

Fig.7 Architecture of TCNN-MobileNet network

图8 11种调制方式的时域图

Fig.8 Time-domain diagrams of 11 modulation modes

图9 小波去噪前后对比

Fig.9 Comparison before and after wavelet denoising

图10 不同Batch size下的识别准确率

Fig.10 Recognition accuracy under different Batch size

图11 不同信噪比下TCNN-MobileNet网络的识别准确率

Fig.11 Recognition accuracy of TCNN-MobileNet network under different signal-to-noise ratio

图12 TCNN-MobileNet在不同Epoch下的训练精度

Fig.12 Training accuracy of TCNN-MobileNet under different Epoch conditions

图13 不同信噪比下的混淆矩阵

Fig.13 Confusion matrix under different signal-to-noise ratio

参考文献 28

1	XU J L, SU W, ZHOU M C. Likelihood-ratio approaches to automatic modulation classification. IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews, 2011, 41(4): 455- 469.
2	XU J L, SU W, ZHOU M C. Software-defined radio equipped with rapid modulation recognition. IEEE Transactions on Vehicular Technology, 2010, 59(4): 1659- 1667. doi: 10.1109/TVT.2010.2041805
3	杨海宇, 郭文普, 康凯. 基于卷积长短时深度神经网络的信号调制方式识别方法. 计算机应用, 2023, 43(4): 1318- 1322. URL
	YANG H Y, GUO W P, KANG K. Signal modulation recognition method based on convolutional long short-term deep neural network. Journal of Computer Applications, 2023, 43(4): 1318- 1322. URL
4	潘一震, 韩顺利, 季桓勇, 等. 基于通道融合的调制信号识别方法. 现代电子技术, 2023, 46(12): 57- 62. URL
	PAN Y Z, HAN S L, JI H Y, et al. Modulation signal recognition method based on channel fusion. Modern Electronics Technique, 2023, 46(12): 57- 62. URL
5	殷赞, 王超杰, 程子恒, 等. 一种基于注意力机制卷积神经网络模型的自动调制识别算法. 电波科学学报, 2023, 38(5): 773- 779. URL
	YIN Z, WANG C J, CHENG Z H, et al. An automatic modulation recognition algorithm based on convolutional neural networks with attention mechanism. Chinese Journal of Radio Science, 2023, 38(5): 773- 779. URL
6	刘高辉, 王壮壮. 基于轻量型网络的单载波信号调制识别. 计算机系统应用, 2023, 32(8): 238- 243. URL
	LIU G H, WANG Z Z. Modulation recognition of single carrier signal based on lightweight network. Computer Systems and Applications, 2023, 32(8): 238- 243. URL
7	任彦洁, 唐晓刚, 张斌权, 等. 基于时间卷积网络的通信信号调制识别算法. 无线电工程, 2023, 53(4): 807- 814. doi: 10.3969/j.issn.1003-3106.2023.04.008
	REN Y J, TANG X G, ZHANG B Q, et al. Communication signal modulation recognition algorithm based on temporal convolutional network. Radio Engineering, 2023, 53(4): 807- 814. doi: 10.3969/j.issn.1003-3106.2023.04.008
8	胡文勇, 童强, 王兵. 一种用于自动调制识别的GR-ResNet方法. 中国无线电, 2023,(5): 65- 69. URL
	HU W Y, TONG Q, WANG B. A GR-ResNet method for automatic modulation identification. China Radio, 2023,(5): 65- 69. URL
9	高锐, 闫光辉, 罗浩, 等. 基于WebSocket技术无线频谱大数据实时监测系统设计与实现. 兰州交通大学学报, 2022, 41(1): 52- 60. URL
	GAO R, YAN G H, LUO H, et al. Design and implementation of a real-time analysis system for big data wireless spectrum monitoring based on WebSocket. Journal of Lanzhou Jiaotong University, 2022, 41(1): 52- 60. URL
10	赵杰, 刘小靖, 胡仁芝. 实时检测与定位电网电压瞬时波动的Coiflets小波方法. 电力系统保护与控制, 2017, 45(15): 8- 14. URL
	ZHAO J, LIU X J, HU R Z. A Coiflets wavelet method for real-timely detecting and locating transient surges of network voltage. Power System Protection and Control, 2017, 45(15): 8- 14. URL
11	牟希农. 离散小波变换在医学影像图像局部压缩中的实现研究. 贵州大学学报(自然科学版), 2021, 38(5): 50- 53. URL
	MOU X N. Research on the realization of discrete wavelet transform in local compression of medical image. Journal of Guizhou University (Natural Sciences), 2021, 38(5): 50- 53. URL
12	刘秋红, 李天昀, 王彬, 等. 基于离散小波分解和频率脊线分析的CPM信号符号速率盲估计. 电子学报, 2020, 48(3): 470- 477. URL
	LIU Q H, LI T Y, WANG B, et al. Symbol rate estimation for CPM signal based on discrete wavelet decomposition and frequency ridge analysis. Acta Electronica Sinica, 2020, 48(3): 470- 477. URL
13	张安康, 吴玉厚, 赵德宏, 等. 基于离散小波分解的刀具故障诊断. 工具技术, 2022, 56(9): 52- 56. URL
	ZHANG A K, WU Y H, ZHAO D H, et al. Tool fault diagnosis based on discrete wavelet decomposition. Tool Engineering, 2022, 56(9): 52- 56. URL
14	谭章禄, 袁慧. 时间序列预处理与信息噪声之间的关系研究-基于离散小波变换和ARIMA模型. 数学的实践与认识, 2020, 50(15): 30- 42. URL
	TAN Z L, YUAN H. Research on the relationship between time series preprocessing and information noise—based on discrete wavelet transform and ARIMA model. Mathematics in Practice and Theory, 2020, 50(15): 30- 42. URL
15	孙洁琪, 李亚峰, 张文博, 等. 基于离散小波变换的双域特征融合深度卷积神经网络. 计算机科学, 2022, 49(S1): 434- 440. URL
	SUN J Q, LI Y F, ZHANG W B, et al. Deep convolutional neural network for dual-domain feature fusion based on discrete wavelet transform. Computer Science, 2022, 49(S1): 434- 440. URL
16	邵婷婷, 白宗文, 周美丽. 基于离散小波变换的信号分解与重构. 计算机技术与发展, 2014, 24(11): 159- 161. URL
	SHAO T T, BAI Z W, ZHOU M L. Decomposition and reconstruction of signal based on DWT. Computer Technology and Development, 2014, 24(11): 159- 161. URL
17	HOWARD A G, ZHU M L, CHEN B, et al. MobileNets: efficient convolutional neural networks for mobile vision applications[EB/OL]. [2023-07-05]. http://arxiv.org/abs/arXiv:1704.04861.
18	ZHANG D N, LU Y Y, LI Y D, et al. Frequency learning attention networks based on deep learning for automatic modulation classification in wireless communication. Pattern Recognition, 2023, 137, 109345.
19	HOU Y M, JIA S Y, LUN X M, et al. Deep feature mining via attention-based BiLSTM-GCN for human motor imagery recognition[EB/OL]. [2023-07-05]. https://arxiv.org/abs/2005.00777v2.
20	WENG L, HE Y, PENG J H, et al. Deep cascading network architecture for robust automatic modulation classification. Neurocomputing, 2021, 455, 308- 324.
21	YI G H, HAO X H, YAN X P, et al. Automatic modulation recognition of radiation source signals based on two-dimensional data matrix and improved residual neural network[EB/OL]. [2023-07-05]. https://www.sciencedirect.com/science/article/pii/S2214914723001885.
22	LIU K, LI F. Automatic modulation recognition based on a multiscale network with statistical features. Physical Communication, 2023, 58, 102052.
23	QI P H, ZHOU X Y, ZHENG S L, et al. Automatic modulation classification based on deep residual networks with multimodal information. IEEE Transactions on Cognitive Communications and Networking, 2021, 7(1): 21- 33.
24	ZHANG Z F, LUO H, WANG C, et al. Automatic modulation classification using CNN-LSTM based dual-stream structure. IEEE Transactions on Vehicular Technology, 2020, 69(11): 13521- 13531.
25	TAN M X, LE Q V. MixConv: mixed depthwise convolutional kernels[EB/OL]. [2023-07-05]. https://arxiv.org/abs/1907.09595v1.
26	TIAN F, WANG Q G, LI X, et al. Heterogeneous multimedia cooperative annotation based on multimodal correlation learning. Journal of Visual Communication and Image Representation, 2019, 58, 544- 553.
27	SANDLER M, HOWARD A, ZHU M L, et al. MobileNetV2: inverted residuals and linear bottlenecks[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2018: 4510-4520.
28	HOWARD A, SANDLER M, CHEN B, et al. Searching for MobileNetV3[C]//Proceedings of IEEE/CVF International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2019: 1314-1324.

[1]	魏嵬, 丁香香, 郭梦星, 杨钊, 刘辉. 文本相似度计算方法综述[J]. 计算机工程, 2024, 50(9): 18-32.
[2]	王志浩, 钱沄涛. 基于Swin Transformer的双流遥感图像时空融合超分辨率重建[J]. 计算机工程, 2024, 50(9): 33-45.
[3]	李俊俊, 董建刚, 李坤. 基于Kubernetes的集群节能策略研究[J]. 计算机工程, 2024, 50(9): 82-91.
[4]	郑秋梅, 赵丹, 牛薇薇, 林超. 基于多通道的彩色图像多重水印算法[J]. 计算机工程, 2024, 50(9): 246-254.
[5]	张鲁, 田春伟, 宋焕生, 刘侍刚. 用于低剂量CT图像去噪的多级双树复小波网络[J]. 计算机工程, 2024, 50(9): 266-275.
[6]	朱凯, 李理, 张彤, 江晟, 别一鸣. 基于Transformer的多阶段运动模糊图像修复网络[J]. 计算机工程, 2024, 50(9): 276-285.
[7]	张天鹏, 韩晶, 吕学强. 基于多任务学习的超分辨率辅助小目标检测[J]. 计算机工程, 2024, 50(9): 304-312.
[8]	高煜宝, 文志诚. 基于注意力机制的双路解码器图像去噪方法[J]. 计算机工程, 2024, 50(9): 324-332.
[9]	张华青, 夏张涛, 陆晓庆, 童基均. 基于字形特征的血管外科命名实体识别[J]. 计算机工程, 2024, 50(8): 13-21.
[10]	王蕾, 党时鹏, 潘丰. 基于卷积神经网络的隐匿性旁路预测模型[J]. 计算机工程, 2024, 50(8): 40-49.
[11]	张亚洲, 和玉, 戎璐, 王祥凯. 基于上下文知识增强型Transformer网络的抑郁检测[J]. 计算机工程, 2024, 50(8): 75-85.
[12]	高伟, 李帅龙, 茆琳, 王磊, 李颖颖, 韩林. 一种基于TVM的算子生成加速策略[J]. 计算机工程, 2024, 50(8): 353-362.
[13]	王宇, 祁琦, 王纯, 许才. 储能变流器信号高精度故障诊断方法[J]. 计算机工程, 2024, 50(8): 389-396.
[14]	耿丽丽, 牛保宁. 基于通道相似度熵的卷积神经网络裁剪[J]. 计算机工程, 2024, 50(7): 133-143.
[15]	张洋, 刘畅, 李少青. 基于可控制性度量的图神经网络门级硬件木马检测方法[J]. 计算机工程, 2024, 50(7): 164-173.

选择文件类型/文献管理软件名称

选择包含的内容