基于1D-Concatenate的信道估计DNN模型优化方法

doi:10.11959/j.issn.1000-0801.2023097

电信科学 ›› 2023, Vol. 39 ›› Issue (4): 71-86.doi: 10.11959/j.issn.1000-0801.2023097

基于1D-Concatenate的信道估计DNN模型优化方法

卢敏¹, 秦泽豪¹^,², 陈志辉¹^,², 张敏¹^,², 乐光学²^,³

¹ 江西理工大学理学院，江西赣州 341000
² 嘉兴学院信息科学与工程学院，浙江嘉兴 314000
³ 浙江省医学电子与数字健康重点实验室，浙江嘉兴 314000

修回日期:2023-04-12 出版日期:2023-04-20 发布日期:2023-04-01
作者简介:卢敏（1964– ），男，江西理工大学理学院教授，主要研究方向为电子材料与器件、智能通信
秦泽豪（1998– ），男，江西理工大学理学院硕士生，嘉兴学院信息科学与工程学院联培研究生，主要研究方向为智能通信、深度学习
陈志辉（1999– ），男，江西理工大学理学院硕士生，嘉兴学院信息科学与工程学院联培研究生，主要研究方向为人工智能、深度学习
张敏（1996– ），女，江西理工大学理学院硕士生，嘉兴学院信息科学与工程学院联培研究生，主要研究方向为深度学习与智能通信
乐光学（1963– ），男，博士，嘉兴学院信息科学与工程学院、浙江省医学电子与数字健康重点实验室教授，主要研究方向为多云融合与协同服务、边缘计算与一体化通信网络、深度学习与智能通信
基金资助:
国家自然科学基金重点项目(U19B2015)

1D-Concatenate based channel estimation DNN model optimization method

Min LU¹, Zehao QIN¹^,², Zhihui CHEN¹^,², Min ZHANG¹^,², Guangxue YUE²^,³

¹ College of Science, Jiangxi University of Technology, Ganzhou 341000, China
² College of Information Science and Engineering, Jiaxing University, Jiaxing 314000, China
³ Zhejiang Key Laboratory of Medical Electronics and Digital Health, Jiaxing 314000, China

Revised:2023-04-12 Online:2023-04-20 Published:2023-04-01
Supported by:
The National Natural Science Foundation Key Project of China(U19B2015)

摘要/Abstract

摘要：

为提高DNN模型在无线通信中信道估计精度，提出一种基于1D-Concatenate的信道估计DNN模型优化方法。该方法将Concatenate进行一维（1D）数据转换，以跳跃连接的方式引入DNN模型，抑制梯度消失问题，运用1D-Concatenate恢复网络训练过程中丢失的数据特征，提高DNN信道估计精度。为验证优化方法的有效性，选取较典型的基于DNN的无线通信信道估计模型进行对比仿真实验。实验结果表明，本文提出的优化方法对已有DNN模型的估计增益提升可达77.10%，在高信噪比下信道增益提升可达3 dB。该优化方法能有效提高DNN模型在无线通信中的信道估计精度，特别是高信噪比下提升效果显著。

关键词: 信道估计, 深度神经网络, Concatenate维度转换, 数据特征恢复

Abstract:

In order to improve the channel estimation accuracy of DNN model in wireless communication, a DNN model optimization method based on 1D-Concatenate was proposed.In this method, Concatenate performs one-dimensional data transformation, the DNN model was introduced by hopping connection, the gradient disappearance problem was suppressed, and 1D-Concatenate was used to restore the data features lost during network training to improve the accuracy of DNN channel estimation.In order to verify the effectiveness of the optimization method, a typical DNN-based wireless communication channel estimation model was selected for comparative simulation experiments.Experimental results show that the estimated gain of the existing DNN model can be increased by 77.10% by the proposed optimization method, and the channel gain can be increased by up to 3 dB under high signal-to-noise ratio.This optimization method can effectively improve the channel estimation accuracy of DNN model in wireless communication, especially the improvement effect is significant under high signal-to-noise ratio.

Key words: channel estimation, deep neural network, Concatenate dimension conversion, data feature recovery

中图分类号:

TN393

卢敏, 秦泽豪, 陈志辉, 张敏, 乐光学. 基于1D-Concatenate的信道估计DNN模型优化方法[J]. 电信科学, 2023, 39(4): 71-86.

Min LU, Zehao QIN, Zhihui CHEN, Min ZHANG, Guangxue YUE. 1D-Concatenate based channel estimation DNN model optimization method[J]. Telecommunications Science, 2023, 39(4): 71-86.

图/表 17

图1

图2

图3

图4

图5

图6

表1

图7

表2

表3

表4

图8

表5

图9

图10

图11

图12

参考文献 21

[22]	JIANG R K , WANG X T , CAO S ,et al. Deep neural networks for channel estimation in underwater acoustic OFDM systems[J]. IEEE Access, 2019(7): 23579-23594.
[23]	BASODI S , JI C Y , ZHANG H P ,et al. Gradient amplification:an efficient way to train deep neural networks[J]. Big Data Mining and Analytics, 2020,3(3): 196-207.
[1]	ZHANG Z F , WANG C , GAN C Q ,et al. Automatic modulation classification using convolutional neural network with features fusion of SPWVD and BJD[J]. IEEE Transactions on Signal and Information Processing Over Networks, 2019,5(3): 469-478.
[2]	廖勇, 花远肖, 姚海梅 ,等. 高速移动环境下基于深度学习的信道估计方法[J]. 电子学报, 2019,47(8): 1701-1707.
	LIAO Y , HUA Y X , YAO H M ,et al. Channel estimation method based on deep learning in high-speed mobile environments[J]. Acta Electronica Sinica, 2019,47(8): 1701-1707.
[3]	张继东, 郑宝玉 . 基于导频的OFDM信道估计及其研究进展[J]. 通信学报, 2003,24(11): 116-124.
	ZHANG J D , ZHENG B Y . Overview of pilot-aided channel estimation in OFDM[J]. Journal of China Institute of Communications, 2003,24(11): 116-124.
[4]	华郁秀, 李荣鹏, 赵志峰 ,等. 基于生成对抗网络的 MIMO信道估计方法[J]. 电信科学, 2021,37(6): 14-22.
	HUA Y X , LI R P , ZHAO Z F ,et al. MIMO channel estimation method based on generating countermeasure network[J]. Telecommunications Science, 2021,37(6): 14-22.
[5]	HUYNH-THE T , HUA C H , NGO T T ,et al. Image representation of pose-transition feature for 3D skeleton-based action recognition[J]. Information Sciences, 2020(513): 112-126.
[6]	PURWINS H , LI B , VIRTANEN T ,et al. Deep learning for audio signal processing[J]. IEEE Journal of Selected Topics in Signal Processing, 2019,13(2): 206-219.
[7]	IQBAL A , THAM M L , CHANG Y C . Double deep Q-network for power allocation in cloud radio access network[C]// Proceedings of 2020 IEEE 3rd International Conference on Computer and Communication Engineering Technology (CCET). Piscataway:IEEE Press, 2020: 272-277.
[8]	IQBAL A , THAM M L , CHANG Y C . Double deep Q-network-based energy-efficient resource allocation in cloud radio access network[J]. IEEE Access, 2021(9): 20440-20449.
[9]	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, 2018,7(1): 114-117.
[10]	陈建廷, 向阳 . 深度神经网络训练中梯度不稳定现象研究综述[J]. 软件学报, 2018,29(7): 2071-2091.
	CHEN J T , XIANG Y . Survey of unstable gradients in deep neural network training[J]. Journal of Software, 2018,29(7): 2071-2091.
[11]	GAO X X , JIN S , WEN C K ,et al. ComNet:combination of deep learning and expert knowledge in OFDM receivers[J]. IEEE Communications Letters, 2018,22(12): 2627-2630.
[12]	BAO Y C , TAN Z Y , SUN H F ,et al. SimNet:simplified deep neural networks for OFDM channel estimation[C]// Proceedings of 2020 IEEE 3rd International Conference on Information Communication and Signal Processing (ICICSP). Piscataway:IEEE Press, 2020: 348-352.
[13]	YANG Y W , GAO F F , MA X L ,et al. Deep learning-based channel estimation for doubly selective fading channels[J]. IEEE Access, 2019(7): 36579-36589.
[14]	SOLTANI M , POURAHMADI V , MIRZAEI A ,et al. Deep learning-based channel estimation[J]. IEEE Communications Letters, 2019,23(4): 652-655.
[15]	SOLTANI M , POURAHMADI V , SHEIKHZADEH H . Pilot pattern design for deep learning-based channel estimation in OFDM systems[J]. IEEE Wireless Communications Letters, 2020,9(12): 2173-2176.
[16]	CHENG C H , HUANG Y F , CHEN H C ,et al. Neural network-based estimation for OFDM channels[C]// Proceedings of 2015 IEEE 29th International Conference on Advanced Information Networking and Applications. Piscataway:IEEE Press, 2015: 600-604.
[17]	BAI Q B , WANG J T , ZHANG Y ,et al. Deep learning-based channel estimation algorithm over time selective fading channels[J]. IEEE Transactions on Cognitive Communications and Networking, 2020,6(1): 125-134.
[18]	唐昊, 韩东洪, 周蕊 . 浅析 OFDM 技术及其应用[J]. 网络安全技术与应用, 2019(5): 62-64.
	TANG H , HAN D H , ZHOU R . Analysis of OFDM technology and its application[J]. Network Security Technology ＆ Application, 2019(5): 62-64.
[19]	LI L J , CHEN H , CHANG H H ,et al. Deep residual learning meets OFDM channel estimation[J]. IEEE Wireless Communications Letters, 2020,9(5): 615-618.
[20]	HUANG G , LIU Z , VAN DER MAATEN L ,et al. Densely connected convolutional networks[C]// Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2017: 2261-2269.
[21]	ZHAO Z Y , VURAN M C , GUO F J ,et al. Deep-waveform:a learned OFDM receiver based on deep complex-valued convolutional networks[J]. IEEE Journal on Selected Areas in Communications, 2021,39(8): 2407-2420.

网络层名称	网络类型	输出大小
Input	Feature Input	(54,1)
FC 1	Fully Connected	Size of baseline model
FC 2	Fully Connected	Size of baseline model
…	…	…
…	…	…
…	…	…
FC L-1	Fully Connected	Size of baseline model
1D-Concat	Concat[FC 2, FC L-1]	(Size of(FC 2+FC L-1),1)
FC L	Fully Connected	（108,1）
Output	Regression Output	Regression

发射端参数		衰落信道参数
名称	值	名称	值
子载波数	64	路径时延	[0.0 0.5 1.2]×10^-6
导频数	27	平均路径增益	[0.1 0.5 0.2]
符号数	4	直射路径多普勒偏移	5.0
发射帧数	1 000	直射路径初始相位	0.5
保护子载波数	4	最大多普勒频移	50
调制方式	BPSK	信道类型	Rice+AWGN

基准模型	优化方法	改进模型
FC-DNN	基于1D-Concatenate的	DenseFC-DNN
SimNet	信道估计 DNN 模型	DenseSimNet
DL-based	优化方法	DenseDL-based
LS_RefineNet		DenseLS_RefineNet
DNN-1		DenseDNN-1
DNN-2		DenseDNN-2

模型名称	数据名称				数值
DenseSimNet	初始学习率	0.010	0.015	0.020	0.025	0.030	0.035	0.040
	平均MSE×10²	1.61	1.51	1.48	1.46	1.45	1.45	1.45
DenseFC-DNN	初始学习率	0.030	0.035	0.040	0.045	0.050	0.055	0.060
	平均MSE×10²	2.48	1.98	1.73	1.74	2.14	1.79	1.78
DenseDL-based	初始学习率	0.010	0.015	0.020	0.025	0.030	0.035	0.040
	平均MSE×10²	2.33	2.03	1.93	1.82	1.86	1.67	1.68
DenseLS_RefineNet	初始学习率	0.010	0.015	0.020	0.025	0.030	0.035	0.040
	平均MSE×10²	1.53	1.46	1.45	1.44	1.44	1.44	1.44
DenseDNN-1	初始学习率	0.005	0.010	0.015	0.020	0.025	0.030	0.035
	平均MSE×10²	3.30	3.00	3.11	3.09	2.67	2.87	2.69
DenseDNN-2	初始学习率	0.005	0.010	0.015	0.020	0.025	0.030	0.035
	平均MSE×10²	2.41	1.55	1.48	1.43	1.38	1.42	1.42

模型名称			SNR
模型名称	1 dB	4 dB	7 dB	10 dB	13 dB
DenseSimNet	2.91	3.31	2.71	3.01	2.84
DenseFC-DNN	2.20	1.91	1.93	1.77	2.10
DenseDL-based	2.01	1.76	1.74	1.81	2.05
DenseLS_RefineNet	1.68	1.46	1.44	1.47	1.60
DenseDNN-1	3.00	3.01	3.16	3.02	3.09
DenseDNN-2	1.81	1.43	1.40	1.44	1.53

基于1D-Concatenate的信道估计DNN模型优化方法

1D-Concatenate based channel estimation DNN model optimization method

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 17

参考文献 21

相关文章 15

Metrics

推荐阅读 0

[1]	李攀攀, 谢正霞, 乐光学, 刘鑫. 基于深度学习的无线通信接收方法研究进展与趋势[J]. 电信科学, 2022, 38(2): 1-17.
[2]	高子路, 孙韶辉, 李丽. 面向新一代移动通信的智能超表面技术综述[J]. 电信科学, 2022, 38(10): 20-35.
[3]	华郁秀, 李荣鹏, 赵志峰, 吴建军, 张宏纲. 基于生成对抗网络的MIMO信道估计方法[J]. 电信科学, 2021, 37(6): 14-22.
[4]	孙姝君, 彭盛亮, 姚育东, 杨喜. 基于深度学习的调制识别综述[J]. 电信科学, 2021, 37(5): 82-90.
[5]	于玄, 耿烜. 一种降低复杂度的压缩感知水声信道估计方法[J]. 电信科学, 2021, 37(3): 114-124.
[6]	程露,杨丽花,王增浩,张捷,梁彦. 基于历史信息的软卡尔曼滤波迭代时变信道估计方法[J]. 电信科学, 2020, 36(9): 23-31.
[7]	邱佳锋. 基于优化BM3D的毫米波大规模MIMO信道估计[J]. 电信科学, 2020, 36(9): 44-50.
[8]	邱佳锋. 宽带毫米波大规模MIMO系统中低复杂度波束域信道估计[J]. 电信科学, 2020, 36(8): 122-129.
[9]	邱佳锋. 基于矩阵完备的低复杂度毫米波大规模MIMO信道估计[J]. 电信科学, 2020, 36(4): 99-106.
[10]	闵锐. 高效深度神经网络综述[J]. 电信科学, 2020, 36(4): 115-124.
[11]	李亚杰,赵永利,刘守东,张杰. 基于人工智能的光纤非线性均衡算法研究概述[J]. 电信科学, 2020, 36(3): 61-70.
[12]	李坤,张静,李潇,金石. 人工智能辅助的信道估计最新研究进展[J]. 电信科学, 2020, 36(10): 46-55.
[13]	刘汉生,唐洪玉,薄明霞,牛剑锋,李天博,李玲晓. 基于机器学习的多源威胁情报质量评价方法[J]. 电信科学, 2020, 36(1): 119-126.
[14]	张瑞,史故臣,刘半藤,陈友荣. 基于韦伯分布函数的低复杂度变步长符号算法[J]. 电信科学, 2018, 34(9): 87-96.
[15]	余翔,高燕妮,段思睿. 多径信道下UFMC系统的信道估计[J]. 电信科学, 2018, 34(4): 22-30.