一种基于RCGAN的水声通信信号降噪方法

doi:10.12263/DZXB.20200369

摘要/Abstract

摘要：

针对复杂海洋环境噪声中水声通信信号数据稀缺条件下的水声通信信号降噪问题，提出一种基于相对条件生成对抗网络（Relativistic Conditional Generative Adversarial Networks，RCGAN）的水声通信信号降噪方法.该方法利用相对条件生成对抗网络具有降噪能力的特点，通过引入扩张卷积结构，构造了适用于水声通信信号的降噪模型，提升了对不同海洋环境噪声的降噪能力；为了解决样本数据稀缺条件下的网络训练问题，根据生成对抗网络特点设计了两步迁移学习策略，并构造了基于数据模型迁移的迁移数据训练集.仿真实验和实际信号测试结果表明，该方法对不同分布特性的海洋环境噪声具有稳健性，能够大幅度降低对目标信号训练数据数量的要求，降噪效果优于现有算法.

关键词: 水声通信信号, 降噪, RCGAN, 扩张卷积, 迁移学习

Abstract:

Aiming at the problem of underwater acoustic communication(UWAC) signal denoising under the conditions of complex marine ambient noise and scarce data, an approach based on relativistic conditional generative adversarial networks(RCGAN) is proposed. Based on the noise reduction capability of the RCGAN, this approach constructs a noise reduction model by introducing the structure of dilated convolution. To overcome the insufficient training data, a two-step transfer learning strategy is designed according to the characteristics of the RCGAN, and a transfer data training set is built based on the presented transfer data model. The results of simulation experiments and practical signal tests both demonstrate the robustness of the proposed method to complex marine environments. Furthermore, the requirements for target signal training data is significantly reduced, with a better performance.

Key words: underwater acoustic communication signal, noise reduction, relativistic conditional generative adversarial networks(RCGAN), dilated convolution, transfer learning

中图分类号:

TN911.7

李勇斌, 王彬, 邵高平, 邵帅. 一种基于RCGAN的水声通信信号降噪方法[J]. 电子学报, 2022, 50(1): 54-62.

LI Yong-bin, WANG Bin, SHAO Gao-ping, SHAO Shuai. A Method of Noise Reduction for Underwater Acoustic Communication Signal Based on RCGAN[J]. Acta Electronica Sinica, 2022, 50(1): 54-62.

图/表 15

图1 基于RCGAN的水声通信信号降噪模型

图2 生成器网络结构

图3 判决器网络结构

图4 RCGAN模型迁移两步训练方法示意图

图5 信号数据结构

图6 BPSK信号降噪前后时域波形和功率谱图对比

图7 8FSK信号降噪前后时频图对比

图8 高斯噪声下不同方法降噪性能对比

图9 Alpha稳定分布噪声下不同方法降噪性能对比

图10 微调不同网络层参数降噪性能对比

图11 不同微调数据量降噪性能对比

表1 实验参数

水域	发送深度	接收深度	水域深度	距离
校园人工湖	0.8 m	0.8 m	1~3 m	50 m
厦门五缘湾	3 m	3 m	6~8 m	545 m

图12 BPSK信号降噪前后时域波形和功率谱图对比

图13 厦门五缘湾海域实验设备布放情况

图14 BPSK信号降噪前后时域波形和功率谱图对比

参考文献 23

1	李亚安, 王洪超, 陈静. 基于奇异谱分解的水声信号降噪方法研究[J]. 系统工程与电子技术, 2007, 29(4): 524-527.
	LI Y A, WANG H C, CHEN J. Research of noise reduction of underwater acoustic signals based on singular spectrum analysis[J]. Systems Engineering and Electronics, 2007, 29(4): 524-527. (in Chinese)
2	WEISS L G, DIXON T L. Wavelet-based denoising of underwater acoustic signals[J]. The Journal of the Acoustical Society of America, 1997, 101(1): 377-383.
3	王剑, 梁斌, 薛飞. 一种基于白噪声EMD分解统计特性的水声信号去噪方法[J]. 四川兵工学报, 2015, 36(6): 131-134.
	WANG J, LIANG B, XUE F. Method of underwater acoustic signal denoising based on characteristics of empirical mode decomposition of white noise[J]. Journal of Sichuan Ordnance, 2015, 36(6): 131-134. (in Chinese)
4	周士贞, 曾向阳, 王璐. 水声信号动态阈值正交匹配追踪降噪方法[J]. 声学技术, 2017, 36(4): 378-382.
	ZHOU S Z, ZENG X Y, WANG L. Dynamic threshold orthogonal matching pursuit method for underwater acoustic signal de-noising[J]. Technical Acoustics, 2017, 36(4): 378-382. (in Chinese)
5	BARAZIDEH R, SUN W S, NATARAJAN B, et al. Impulsive noise mitigation in underwater acoustic communication systems: Experimental studies[C]//2019 IEEE 9th Annual Computing and Communication Workshop and Conference (CCWC). Piscataway, USA: IEEE, 2019： 880-885.
6	CHITRE M, POTTER J, HENG O S. Underwater acoustic channel characterisation for medium-range shallow water communications[C]//Oceans'04 MTS/IEEE Techno-Ocean'04 (IEEECat. No.04CH37600). Piscataway, USA: IEEE, 2004: 40-45.
7	MIKOSCH T, SAMORODNITSKY G, TAQQU M S. Stable non-Gaussian random processes: Stochastic models with infinite variance[J]. Journal of the American Statistical Association, 1995, 90(430): 805.
8	BABY D, VERHULST S. Sergan: Speech enhancement using relativistic generative adversarial networks with gradient penalty[C]//ICASSP 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Piscataway, USA: IEEE, 2019: 106-110.
9	袁文浩, 娄迎曦, 梁春燕, 等. 利用生成噪声提高语音增强方法的泛化能力[J]. 电子学报, 2019, 47(4): 791-797.
	YUAN W H, LOU Y X, LIANG C Y, et al. Improving generalization ability of speech enhancement approaches using generated noise[J]. Acta Electronica Sinica, 2019, 47(4): 791-797. (in Chinese)
10	GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[C]//Advances in Neural Information Processing Systems (NIPS). Montreal, CA: MIT Press, 2014: 2672-2680.
11	MIRZA M, OSINDERO S. Conditional generative adversarial nets[EB/OL]. [2021]. .
12	YU F, KOLTUN V. Multi-scale context aggregation by dilated convolutions[C]//Proceedings of the International Conference on Learning Representations (ICLR). New York, USA: ACM, 2016: 1-13.
13	RADFORD A, METZ L, CHINTALA S. Unsupervised representation learning with deep convolutional generative adversarial networks[EB/OL]. [2021]. .
14	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, USA: IEEE, 2016: 770-778.
15	权宇, 李志欣, 张灿龙, 等. 融合深度扩张网络和轻量化网络的目标检测模型[J]. 电子学报, 2020, 48(2): 390-397.
	QUAN Y, LI Z X, ZHANG C L, et al. Fusing deep dilated convolutions network and light-weight network for object detection[J]. Acta Electronica Sinica, 2020, 48(2): 390-397. (in Chinese)
16	MAO X D, LI Q, XIE H R, et al. Least squares generative adversarial networks[C]//2017 IEEE International Conference on Computer Vision (ICCV). Piscataway, USA: IEEE, 2017: 2813-2821.
17	ARJOYSKY M, CHINTALA S, BOTTOU L. Wasserstein generative adversarial networks[C]//Proceedings of the International Conference on Machine Learning (ICML). New York, USA: ACM, 2017: 214-223.
18	GULRAJANI I, AHMED F, ARJOYSKY M, et al. Improved training of wasserstein gans[C]//Advances in Neural Information Processing Systems (NIPS). Montreal, CA: MIT Press, 2017: 5767-5777.
19	JOLICOEUR-MARTINEAU A. The relativistic discriminator: A key element missing from standard GAN[C]//Proceedings of the International Conference on Learning Representations (ICLR). New York, USA: ACM, 2019: 1-26.
20	YOSINSKI J, CLUNE J, BENGIO Y, et al. How transferable are features in deep neural networks?[C]//Advances in Neural Information Processing Systems (NIPS). Montreal, CA: MIT Press, 2014: 3320-3328.
21	肖瑛. 水声信道常数模盲均衡: 理论、算法与仿真[M]. 北京: 人民邮电出版社, 2015.
22	ZENG C Z, JIA X. Modulation recognition method of communication signals based on correlation characteristics[C]//2015 IEEE International Conference on Signal Processing, Communications and Computing (ICSPCC). Piscataway, USA: IEEE, 2015: 1-5.
23	MCCULLOCH J H. Simple consistent estimators of stable distribution parameters[J]. Communications in Statistics - Simulation and Computation, 1986, 15(4): 1109-1136.

[1]	李维刚, 甘平, 谢璐, 李松涛. 基于样本对元学习的小样本图像分类方法[J]. 电子学报, 2022, 50(2): 295-304.
[2]	张小恒, 张馨月, 李勇明, 王品, 刘玉川. 面向帕金森病语音诊断的非监督两步式卷积稀疏迁移学习算法[J]. 电子学报, 2022, 50(1): 177-184.
[3]	吴龙文, 聂雨亭, 张宇鹏, 何胜阳, 赵雅琴. 基于变分模态分解的自适应滤波降噪方法[J]. 电子学报, 2021, 49(8): 1457-1465.
[4]	唐焕玲, 郑涵, 刘艳红, 马思源, 窦全胜, 鲁明羽. Tr-SLDA:一种面向交叉领域的迁移主题模型[J]. 电子学报, 2021, 49(3): 605-613.
[5]	江泽涛, 钱艺, 伍旭, 张少钦. 一种基于ARD‑GAN的低照度图像增强方法[J]. 电子学报, 2021, 49(11): 2160-2165.
[6]	孙鑫威, 纪爱敏, 陈曦晖, 林新海, 许行. 强噪声背景下动车组轴承微弱故障信号检测[J]. 电子学报, 2021, 49(11): 2217-2224.
[7]	张锦, 李阳, 任传伦, 黄炼, 王帅辉, 段晔鑫, 潘志松, 谢钧. 基于帧间高级特征差分的跨场景视频前景分割算法[J]. 电子学报, 2021, 49(10): 2032-2040.
[8]	王格格, 郭涛, 余游, 苏菡. 基于生成对抗网络的无监督域适应分类模型[J]. 电子学报, 2020, 48(6): 1190-1197.
[9]	李东瑾, 杨瑞娟, 李晓柏, 董睿杰. 基于栈式稀疏降噪自编码网络的辐射源调制识别[J]. 电子学报, 2020, 48(6): 1198-1204.
[10]	张红斌, 蒋子良, 熊其鹏, 武晋鹏, 邬任重, 袁天, 姬东鸿. 基于改进的有效区域基因选择与跨模态语义挖掘的图像属性标注[J]. 电子学报, 2020, 48(4): 790-799.
[11]	赵鹏, 王美玉, 纪霞, 刘慧婷. 基于张量表示的域适配的迁移学习中特征表示方法[J]. 电子学报, 2020, 48(2): 359-368.
[12]	权宇, 李志欣, 张灿龙, 马慧芳. 融合深度扩张网络和轻量化网络的目标检测模型[J]. 电子学报, 2020, 48(2): 390-397.
[13]	刘真, 田靖玉, 苑宝鑫, 孙永奇. 基于知识聚合和迁移的跨领域推荐算法[J]. 电子学报, 2020, 48(10): 1928-1932.
[14]	李庆武, 马云鹏, 周亚琴, 邢俊. 基于无监督栈式降噪自编码网络的显著性检测算法[J]. 电子学报, 2019, 47(4): 871-879.
[15]	季鼎承, 蒋亦樟, 王士同. 基于域与样例平衡的多源迁移学习方法[J]. 电子学报, 2019, 47(3): 692-699.