当前，随着无线通信的发展，OFDM技术作为无线通信中的重要技术被应用于不同场景中。将OFDM技术应用于海洋复杂场景下的无线通信系统中，使得海洋无线通信系统拥有更高的传输速率、更优的通信质量。现有的物理层接收机通常采用载波和符号同步、信道估计、均衡、解调等模块逐步串行的处理方法，预处理模块的误差可能会影响后续处理模块的优化，导致误差累积，从而无法得到总体性能最优的接收端模型。为实现接收端的联合优化，本文尝试使用一个深度神经网络代替OFDM接收端的各个模块。本文研究的内容：

（1）针对OFDM技术现有方法的缺陷以及海洋复杂环境下OFDM信息比特流的特征提取困难、误比特率无法进一步降低的问题，研究基于深度学习的海洋OFDM智能接收方法。使用残差神经网络（ResNet）代替海洋OFDM接收端信道估计、均衡、解调、译码，实现接收端的整体优化。并在ResNet模型中引入通道注意力模块，构建深度神经网络，提取信号特征，最终恢复出信息比特流。分析OFDM系统受导频、载波频偏等因素影响在该模型下的性能，实验结果表明，该模型能够对海洋复杂场景下的OFDM信号特征进行有效的提取，实现较低的误比特率。

（2）针对OFDM系统中存在很高的峰值-均值功率比（PAPR），导致通过放大器后的OFDM信号更易产生非线性失真，使得子载波之间的正交性遭到破坏，进而恶化传输性能的问题，研究基于神经网络抑制PAPR的OFDM智能接收方法。在OFDM系统发射端的串/并转换和IFFT之间使用一个的多层神经网络来抑制PAPR，并与提出的OFDM智能接收模型联合，旨在使OFDM系统误比特率的增长在可控范围内的情况下抑制PAPR。实验结果表明，该模型在抑制PAPR的方面具有显著的效果，并且其误比特率的增长在可控范围内。

At present, with the development of wireless communication, OFDM technology is used in different scenarios as an important technology in wireless communication. The application of OFDM technology to wireless communication systems in complex marine scenarios enables marine wireless communication systems to have higher transmission rates and better communication quality. Existing physical layer receivers usually use progressive serial processing of modules such as carrier and symbol synchronization, channel estimation, equalization, demodulation, etc. The error of the preprocessing module may affect the optimization of the subsequent processing module, resulting in the accumulation of errors, so the receiver model with the best overall performance cannot be obtained. In order to realize the joint optimization of the receiver, this thesis attempts to use a deep neural network to replace the various modules of the OFDM receiver. The research contents of this thesis as follows:

 (1) Aiming at the defects of the existing methods of OFDM technology and the difficulty of feature extraction of OFDM information bit stream in the complex marine environment, and the problem that the bit error rate cannot be further reduced, an intelligent receiving method for OFDM in ocean based on deep learing is studied. A ResNet is used to replace the channel estimation, equalization, demodulation and decoding at the receiver of marine OFDM to realize the overall optimization of the receiver. And the channel attention module is introduced into the ResNet model to build a deep neural network, extract signal features, and finally restore the information bit stream. The performance of the OFDM system affected by the pilot frequency, carrier frequency offset and other factors under the model is analyzed. The experimental results show that the model can effectively extract the characteristics of the OFDM signal in the complex marine scene and achieve a lower bit error rate.

(2) Aiming at the high PAPR in OFDM system, which destroys the orthogonality between subcarriers and worsens the transmission performance, an OFDM intelligent receiving method based on neural network to suppress PAPR is studied. A multilayer neural network is used between the serial to parallel conversion and IFFT at the transmitter of OFDM system to suppress PAPR, and combined with the proposed OFDM intelligent receiving model to suppress PAPR when the bit error rate of OFDM system increases within a controllable range. The experimental results show that the model has a significant effect in suppressing PAPR, and the increase of bit error rate is controllable.

[1] 于永学, 王玉珏, 解嘉宇. 海洋通信的发展现状及应用构想[J]. 海洋信息, 2020, 35(02): 25-28.

[2] 夏明华, 朱又敏等. 海洋通信的发展现状与时代挑战[J]. 中国科学: 信息科学, 2017, 47(06): 677-695.

[3] 金鹏, 熊瑾煜, 张晓勇. 海上无线电监测浮标的覆盖范围研究[J]. 舰船电子工程, 2017, 37(05): 20-24+104.

[4] 仪青帝. 海域电磁波传播模型研究[D]. 海南: 海南大学, 2015.

[5] 李彪. 基于Longley-Rice模型的海上无线电通信时间分析[J]. 科技创新与应用, 2018, 13: 16-17.

[6] 邓力, 俞栋. 深度学习: 方法及应用[M]. 谢磊, 译. 北京: 机械工业出版社, 2016: 197-387.

[7] Wang T, Wen C K, Wang H, et al. Deep learning for wireless physical layer: opportunities and challenges[J]. 中国通信: 英文版, 2017, 14(11):20.

[8] Qiao G, Babar Z, Ma L, et al. MIMO-OFDM underwater acoustic communication systems—A review[J]. Physical Communication, 2017, 23: 56-64.

[9] 杨宏晖, 申昇, 姚晓辉等. 用于水声目标特征学习与识别的混合正则化深度置信网络[J]. 西北工业大学学报, 2017, 35(02): 220-225.

[10] Qiao G, Babar Z, Ma L, et al. Shallow water acoustic channel modeling and MIMO-OFDM simulations[C]//2018 15th international bhurban conference on applied sciences and technology (IBCAST). IEEE, 2018: 709-715.

[11] Ramadan K, Dessouky M I, Elagooz S, et al. Equalization and carrier frequency offset compensation for underwater acoustic OFDM systems[J]. Annals of Data Science, 2018, 5(02): 259-272.

[12] Zhang Y, Li C, Wang H, et al. Deep learning aided OFDM receiver for underwater acoustic communications[J]. Applied Acoustics, 2022, 187: 108515.

[13] Jiang R, Wang X, Cao S, et al. Deep neural networks for channel estimation in underwater acoustic OFDM systems[J]. IEEE Access, 2019, 7: 23579 - 23594.

[14] Mao Q, Hu F, Hao Q. Deep learning for intelligent wireless networks: A comprehensive survey[J]. IEEE Communications Surveys & Tutorials, 2018, 20(04): 2595-2621.

[15] Qin Z, Ye H, Li G Y, et al. Deep learning in physical layer communications[J]. IEEE Wireless Communications, 2019, 26(02): 93-99.

[16] KIM M, LEE W, CHO D H. A novel PAPR reduction scheme for OFDM system based on deep learning[J]. IEEE Communications Letters, 2018, 22 (03):510-513.

[17] 陈嘉润, 余宝贤, 王剑莹等. 基于深度神经网络的高频谱效率频分复用系统的信道估计方法[J]. 华南师范大学学报(自然科学版), 2020, 52(03): 17-21.

[18] 廖勇, 花远肖, 姚海梅. 基于深度学习的OFDM信道估计[J]. 重庆邮电大学学报(自然科学版), 2019, 31(03): 348-353

[19] Ye H, Li G Y, Juang B H. Power of deep learning for channel estimation and signal detection in OFDM systems[J]. IEEE Wireless Communications Letters, 2017, 7(01): 114-117.

[20] Ha C B, Song H K. Signal detection scheme based on adaptive ensemble deep learning model[J]. IEEE Access, 2018, 6: 21342-21349.

[21] 常代娜, 周杰. 基于深度学习算法的OFDM信号检测[J]. 东南大学学报(自然科学版), 2020, 50(05): 912-917.

[22] 陈晋音, 成凯回, 郑海斌. 低信噪比下基于深度学习的调制模式识别方法[J]. 计算机科学, 2020, 47(S1): 283-288.

[23] 徐建业, 杨霄鹏, 李伟等. 基于动态学习率深度神经网络的抗干扰信道编码算法[J]. 计算机应用研究, 2020, 37(07): 2171-2174.

[24] Oshea T, Hoydis J. An introduction to deep learning for the physical layer[J]. IEEE Transactions on Cognitive Communications & Networking, 2017, 3(04):563-575.

[25] Dorner S, Cammerer S, Hoydis J, et al. On deep learning-based communication over the air[C]// 2017 51st Asilomar Conference on Signals, Systems, and Computers. IEEE, 2018.

[26] O'Shea T J, Karra K, Clancy T C. Learning to communicate: Channel auto-encoders, domain specific regularizers, and attention[C]//2016 IEEE International Symposium on Signal Processing and Information Technology (ISSPIT). IEEE, 2016: 223-228.

[27] Huang H, Yang J, Huang H, et al. Deep learning for super-resolution channel estimation and DOA estimation based massive MIMO system[J]. IEEE Transactions on Vehicular Technology, 2018, 67(09): 8549-8560.

[28] Felix A, Cammerer S, Dörner S, et al. OFDM-autoencoder for end-to-end learning of communications systems[C]//2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC). IEEE, 2018: 1-5.

[29] Ye H, Li G Y, Juang B H F, et al. Channel agnostic end-to-end learning based communication systems with conditional GAN[C]//2018 IEEE Globecom Workshops (GC Wkshps). IEEE, 2018: 1-5.

[30] 赵勇洙, 金宰权, 杨元永. MIMO-OFDM无线通信技术及MATLAB仿真[M]. 孙锴, 译. 北京: 电子工业出版社, 2012: 175-206.

[31] 袁建国, 李尚晋, 赵富强等. 可见光OFDM系统中峰均比抑制技术的研究进展[J]. 激光杂志, 2020, 41(10): 1-7.

[32] Zhitong Xing, Kaiming Liu, Kaiyuan Huang, et al. Novel PAPR reduction scheme based on continuous nonlinear piecewise companding transform for OFDM systems[J]. 中国通信, 2020, 17(09): 177-192.

[33] 邢智童, 李云, 彭德义等. OFDM中一种有效的基于分段非线性压扩的PAPR抑制算法[J]. 通信学报, 2021, 42(12): 44-53.

[34] 吴飞腾, 王军, 宋健. 降低OFDM系统带外泄漏和PAPR的新预编码方法[J]. 电讯技术, 2018, 58(12): 1421-1426.

[35] Zhang W, Hajomer A, Yang X, et al. Neural networks for PAPR reduction in optical OFDM signal transmission[C]//2019 21st International Conference on Transparent Optical Networks (ICTON). IEEE, 2019: 1-4.

[36] Insoo Sohn. A low complexity PAPR reduction scheme for OFDM systems via neural networks[J]. IEEE Communications Letters, 2014, 18(02): 225-228.

[37] 魏特, 王文浩, 陈军等. 环境信息辅助的海上无线信道测量与建模[J]. 清华大学学报(自然科学版), 2021, 61(09): 1002-1007

[38] 兰田, 全厚德, 崔佩璋等. 起伏地形下超短波电波传播模型修正方法[J]. 计算机仿真, 2019, 36(10): 148-152.

[39] Lee J H, Choi J, Lee W H, et al. Measurement and analysis on land-to-ship offshore wireless channel in 2.4 GHz[J]. IEEE Wireless Communications Letters, 2017, 6(02): 222-225.

[40] 佟禹瑄, 林彬, 任鸿翔等. 海上无线传输虚拟可视化仿真平台设计与实现[J]. 系统仿真学报, 2018, 30(11): 4227-4235.

[41] 樊昌信, 曹丽娜. 通信原理(第7版)[M]. 北京: 电子工业出版社, 2018: 242-248.

[42] 李林. 基于深度学习的物理层接收机模型设计[D]. 四川: 西南交通大学,2019.

[43] Zheng S, Chen S, Yang X. Deep receiver: a deep learning-based intelligent receiver for wireless communications in the physical layer[J]. IEEE Transactions on Cognitive Communications and Networking, 2020, 7(01): 5-20.

[44] Lecun Y, Boser B, Denker J, et al. Backpropagation applied to handwritten zip code recognition[J]. Neural Computation, 2014, 1(04): 541-551.

[45] 魏明珠, 郑荣, 杨竞雄. 基于深度学习的图像检索研究进展[J]. 情报科学, 2021, 39 (05): 184-192.

[46] 俞颂华. 卷积神经网络的发展与应用综述[J]. 信息通信, 2019, 2: 39-43.

[47] 严春满, 王铖. 卷积神经网络模型发展及应用[J]. 计算机科学与探索, 2021, 15(01): 27-46.

[48] 林钰棽, 魏云龙, 陈琪琪等. 一种嵌入式计算平台的Sigmoid函数优化方法[J]. 小型微型计算机系统, 2021, 42(10): 2053-2058.

[49] 张博文, 陈刚, 陈旭等. 基于二阶近似和误差补偿的Tanh函数硬件实现[J]. 微电子学, 2021, 51(06): 905-909.

[50] 蒋昂波, 王维维, JIANG等. ReLU激活函数优化研究[J]. 传感器与微系统, 2018, 37(02): 3.

[51] 张玉宏. 深度学习之美[M]. 北京: 电子工业出版社, 2021: 88-129.

[52] 弗朗索瓦·肖莱. Python深度学习[M]. 张亮, 译. 北京: 人民邮电出版社, 2019: 136-157.

[53] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]// Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.

[54] 张宸嘉, 朱磊, 俞璐. 卷积神经网络中的注意力机制综述[J]. 计算机工程与应用, 2021, 57(20): 64-72.

[55] Woo S, Park J, Lee J Y, et al. CBAM: Convolutional block attention module[C]// Proceedings of the European conference on computer vision (ECCV). 2018: 3-19.

[56] Jie H, Li S, Gang S, et al. Squeeze-and-excitation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020 42(08): 2011-2023.

[57] 袁田, 朱红亮, 周娟等. 降低OFDM立方度量的最优限幅滤波算法及神经网络实现[J]. 计算机工程与科学, 2020, 42(08): 1359-1366.