Please wait a minute...
浙江大学学报(工学版)  2020, Vol. 54 Issue (4): 732-738    DOI: 10.3785/j.issn.1008-973X.2020.04.012
计算机技术、信息工程     
基于深度学习的多载波系统信道估计与检测
汪周飞(),袁伟娜*()
华东理工大学 信息科学与工程学院,上海 200237
Channel estimation and detection method for multicarrier system based on deep learning
Zhou-fei WANG(),Wei-na YUAN*()
School of Information Science and Engineering, East China University of Science and Technology, Shanghai 200237, China
 全文: PDF(1007 KB)   HTML
摘要:

为了提升滤波器组多载波(FBMC)系统的通信质量,针对符号检测与信道估计问题,研究系统框架和虚部干扰问题,提出基于深度学习的FBMC系统信道估计与检测方法. 搭建完整的FBMC-偏移正交幅度调制(OQAM)系统与深度学习模型结合的仿真系统,设计接收数据的特征与标签处理;采用ResNet-DNN神经网络对信道符号检测模块建模,改进原模型网络结构和优化模型参数,和传统的分类器相比,提高了符号检测的准确性;采用CNN+NN模型对信道估计、均衡、符号检测模块进行建模和集成,理论分析和仿真结果表明,新方法的抗噪声能力、鲁棒性和误比特率(BER)性能均优于正交频分复用(OFDM)系统和基于导频估计的FBMC系统性能.

关键词: 信道估计滤波器组多载波(FBMC)深度学习神经网络符号检测    
Abstract:

The system framework and imaginary interference were analyzed aiming at the problem of symbol detection and channel estimation in order to effectively improve the communication quality of filter bank multicarrier (FBMC) system. A channel estimation and detection method for FBMC system was proposed based on deep learning. A complete simulation system was established by combining FBMC-offset quadrature amplitude modulation (OQAM) with deep learning model, and the characteristics and label processing of received data were designed. ResNet-DNN neural network was used to model the channel symbol detection module. The original model structure and optimized model parameters were improved, which improved the accuracy of symbol detection compared with traditional classifiers. CNN+NN model was used to model and integrate for estimating, equalizing and detecting channel symbols. The theoretical analysis and simulation results show that the new method is superior to orthogonal frequency division multiplexing (OFDM) system and FBMC system based on pilot estimation in terms of noise resistance, robustness and bit error rate (BER) performance.

Key words: channel estimation    filter bank multicarrier (FBMC)    deep learning    neural network    symbol detection
收稿日期: 2019-03-24 出版日期: 2020-04-05
CLC:  TN 929  
基金资助: 国家自然科学基金资助项目(61501187);中央高校基本科研业务费资助项目
通讯作者: 袁伟娜     E-mail: 654007563@qq.com;wnyuan_ice@163.com
作者简介: 汪周飞(1993—),男,硕士生,从事5G、信号处理、人工智能等相关技术研究. orcid.org/0000-0002-9318-5301. E-mail: 654007563@qq.com
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
作者相关文章  
汪周飞
袁伟娜

引用本文:

汪周飞,袁伟娜. 基于深度学习的多载波系统信道估计与检测[J]. 浙江大学学报(工学版), 2020, 54(4): 732-738.

Zhou-fei WANG,Wei-na YUAN. Channel estimation and detection method for multicarrier system based on deep learning. Journal of ZheJiang University (Engineering Science), 2020, 54(4): 732-738.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2020.04.012        http://www.zjujournals.com/eng/CN/Y2020/V54/I4/732

图 1  FBMC-OQAM收发系统框图
步骤 描述
1 接收数据信号虚实分离
2 取实部小数点前2位、小数点后6位数字和1位正负号
3 取虚部小数点前2位、小数点后6位数字和1位正负号
4 采用one-hot编码上述18位数字生成18×12的特征矩阵
表 1  接收数据符号的特征处理步骤
图 2  ResNet-DNN网络模型结构
图 3  符号检测模块流程结构
步骤 步骤说明
1 初始化参数、信道以及FBMC类模块
2 产生随机的原始数据比特流bitstream_FBMC(0和1的序列)
3 bitstream_FBMC经过发送端,具体为符号映射、虚实分离、调制得到发送信号s_FBMC
4 s_FBMC进过信道再加上噪声的干扰得到接收信号r_FBMC
5 r_FBMC经过接收端,解调接收信号,基于导频估计出信道H,乘上1/H得到检测信号xDest_FBMC
6 xDest_FBMC进行特征处理作为数据样本的特征,bitstream_FBM标签处理作为标签,产生训练的数据集X
7 用数据集X进行模型的训练,产生的模型参数存放在model中
8 将model用于检测新的接收信号xDest_FBMC_1,预测新的原始信号计算BER
表 2  符号检测模块流程步骤
参数 数值
QAM阶数 16
子载波间隔 15 kHz
子载波数量 30
FBMC符号数量 15
采样频率 30 kHz
信道名称 ‘AWGN’高斯白信道
载波频率 2.5 GHz
蒙特卡洛迭代次数 20
表 3  FBMC系统的通信仿真参数
参数 数值
激活函数 ReLU
第1个残差层神经元个数 500
后2个残差层神经元个数 200
输出层全连接层神经元个数 16
Dropout率 0.5
损失函数 交叉熵
优化器 Adam
表 4  ResNet-DNN模型仿真参数
图 4  AWGN信道下的BER性能比较
图 5  PedestrianA信道下的BER性能比较
图 6  AWGN信道下增加DFT算法的BER性能比较
图 7  CNN+NN网络模型结构
参数 数值
激活函数 ReLU
第1通道卷积核大小 2×12
第2通道卷积核大小 3×12
正则项 L2正则
Dropout率 0.5
损失函数 交叉熵函数
优化器 Adam
NN隐藏层神经元个数 500
表 5  CNN+NN模型仿真参数
图 8  AWGN信道下的BER性能(CNN+NN)
1 BEAULIEU F D, CHANMPAGNE B. Multicarrier modulation using perfect reconstruction DFT filter bank transceivers [C] // 2005 5th International Conference on Information Communications and Signal Processing. Bangkok: ICICS, 2005: 111-115.
2 VIHOLAINEN A, IHALAINEN T, RENFORS M. Performance of time-frequency localized and frequency selective filter banks in multicarrier systems [C] // 2006 IEEE International Symposium on Circuits and Systems. Island of Kos: IEEE, 2006: 1693494.
3 BEAULIEU F D. Multicarrier transceivers using DFT filter banks with perfect reconstruction [D]. Canada: McGill University, 2007.
4 MARTIN P M, BREGOVIC R, SARAMAKI R. An efficient approach for designing filter banks for multicarrier transmission [C] // 2008 IEEE International Symposium on Circuits and Systems. Seattle, WA: IEEE, 2008: 3206-3209.
5 BOROUJENY B F Filter bank spectrum sensing for cognitive radios[J]. IEEE Transactions on Signal Processing, 2008, 56 (5): 1801- 1811
doi: 10.1109/TSP.2007.911490
6 LELE C, LEGOUABL R, SIOHAN P. Channel estimation with scattered pilots in OFDM/OQAM [C] // 2008 IEEE 9th Workshop on Signal Processing Advances in Wireless Communications. Recife: IEEE, 2008: 286-290.
7 陈孝武. 脉冲成形OFDM/OQAM系统关键技术研究[D]. 杭州: 浙江大学, 2015.
CHEN Xiao-wu. Research on key technologies of pulse forming OFDM/OQAM system [D]. Hangzhou: Zhejiang University, 2015.
8 YE H, LI G Y, JUANG B Power of deep learning for channel estimation and signal detection in OFDM systems[J]. IEEE Wireless Communications Letters, 2018, 7 (1): 114- 117
doi: 10.1109/LWC.2017.2757490
9 YE H, LI G Y. Initial results on deep learning for joint channel equalization and decoding [C] // 2017 IEEE 86th Vehicular Technology Conference. Toronto: IEEE, 2017: 1-5.
10 CHENG X, LIU D, ZHU Z, et al. A ResNet-DNN based channel estimation and equalization scheme in FBMC/OQAM systems [C] // 2018 10th International Conference on Wireless Communications and Signal Processing. Hangzhou: WCSP, 2018: 1-5.
11 CAO Y, ZHAO Y. Text classification based on TextCNN for power grid user fault repairing information [C] // 2018 5th International Conference on Systems and Informatics. Nanjing: ICSAI, 2018: 1182-1187.
12 ZHANG C, ZHANG H, YUAN D, et al Citywide cellular traffic prediction based on densely connected convolutional neural networks[J]. IEEE Communications Letters, 2018, 22 (8): 1656- 1659
doi: 10.1109/LCOMM.2018.2841832
13 JAVAUDIN J P, LACROIX D, ROUXEL A. Pilot-aided channel estimation for OFDM/OQAM [C] // The IEEE 57th IEEE Semiannual Vehicular Technology Conference. South Korea: IEEE, 2003: 1581-1585.
14 LéLé C, LEGOUABLE R, SIOHAN P. Iterative scattered pilot channel estimation in OFDM/OQAM [C] // 2009 IEEE 10th Workshop Signal Process Advances in Wireless Communications. Perugia: IEEE, 2009: 176-180.
[1] 许佳辉,王敬昌,陈岭,吴勇. 基于图神经网络的地表水水质预测模型[J]. 浙江大学学报(工学版), 2021, 55(4): 601-607.
[2] 王虹力,郭斌,刘思聪,刘佳琪,仵允港,於志文. 边端融合的终端情境自适应深度感知模型[J]. 浙江大学学报(工学版), 2021, 55(4): 626-638.
[3] 张腾,蒋鑫龙,陈益强,陈前,米涛免,陈彪. 基于腕部姿态的帕金森病用药后开-关期检测[J]. 浙江大学学报(工学版), 2021, 55(4): 639-647.
[4] 程鸿,胡佳杰,刘勇,叶远青. 强度传输方程和神经网络融合的三维重构算法[J]. 浙江大学学报(工学版), 2021, 55(4): 658-664.
[5] 郑英杰,吴松荣,韦若禹,涂振威,廖进,刘东. 基于目标图像FCM算法的地铁定位点匹配及误报排除方法[J]. 浙江大学学报(工学版), 2021, 55(3): 586-593.
[6] 毛奕喆,龚国芳,周星海,王飞. 基于马尔可夫过程和深度神经网络的TBM围岩识别[J]. 浙江大学学报(工学版), 2021, 55(3): 448-454.
[7] 刘芳,汪震,刘睿迪,王锴. 基于组合损失函数的BP神经网络风力发电短期预测方法[J]. 浙江大学学报(工学版), 2021, 55(3): 594-600.
[8] 徐利锋,黄海帆,丁维龙,范玉雷. 基于改进DenseNet的水果小目标检测[J]. 浙江大学学报(工学版), 2021, 55(2): 377-385.
[9] 陈世达,刘强,韩亮. 降低分布式训练通信的梯度稀疏压缩方法[J]. 浙江大学学报(工学版), 2021, 55(2): 386-394.
[10] 赵燕伟,张健,周仙明,吴耿育. 基于视觉-磁引导的无人机动态跟踪与精准着陆[J]. 浙江大学学报(工学版), 2021, 55(1): 96-108.
[11] 陈纬奇,王敬昌,陈岭,杨勇勤,吴勇. 基于深度神经网络的多因素感知终端换机预测模型[J]. 浙江大学学报(工学版), 2021, 55(1): 109-115.
[12] 许豪灿,李基拓,陆国栋. 由LeNet-5从单张着装图像重建三维人体[J]. 浙江大学学报(工学版), 2021, 55(1): 153-161.
[13] 黄毅鹏,胡冀苏,钱旭升,周志勇,赵文露,马麒,沈钧康,戴亚康. SE-Mask-RCNN:多参数MRI前列腺癌分割方法[J]. 浙江大学学报(工学版), 2021, 55(1): 203-212.
[14] 陈巧红,陈翊,李文书,贾宇波. 多尺度SE-Xception服装图像分类[J]. 浙江大学学报(工学版), 2020, 54(9): 1727-1735.
[15] 郑浦,白宏阳,李伟,郭宏伟. 复杂背景下的小目标检测算法[J]. 浙江大学学报(工学版), 2020, 54(9): 1777-1784.