基于深度学习的广义频分复用系统时频双选择信道估计

doi:10.11959/j.issn.1000-436x.2021188

通信学报 ›› 2021, Vol. 42 ›› Issue (10): 233-242.doi: 10.11959/j.issn.1000-436x.2021188

基于深度学习的广义频分复用系统时频双选择信道估计

王莹¹, 任军¹, 史可¹, 林彬¹^,²

¹ 大连海事大学信息科学技术学院，辽宁大连 116026
² 鹏程实验室网络通信研究中心，广东深圳 518052

修回日期:2021-09-06 出版日期:2021-10-25 发布日期:2021-10-01
作者简介:王莹（1968− ），男，河北保定人，博士，大连海事大学教授，主要研究方向为移动通信理论、无线自组网等
任军（1997− ），男，安徽淮北人，大连海事大学硕士生，主要研究方向为多载波通信技术
史可（1998− ），女，吉林榆树人，大连海事大学硕士生，主要研究方向为多载波通信技术
林彬（1977− ），女，辽宁大连人，博士，大连海事大学教授，主要研究方向为无线网络优化理论
基金资助:
国家重点研发计划基金资助项目(2019YFE0111600);国家自然科学基金资助项目(61971083);国家自然科学基金资助项目(51939001);大连市科技创新基金资助项目(2019J11CY015)

Doubly-selective channel estimation for generalized frequency division multiplexing systems based on deep learning

Ying WANG¹, Jun REN¹, Ke SHI¹, Bin LIN¹^,²

¹ College of Information Science and Technology, Dalian Maritime University, Dalian 116026, China
² Research Center of Network Communications, Peng Cheng Laboratory, Shenzhen 518052, China

Revised:2021-09-06 Online:2021-10-25 Published:2021-10-01
Supported by:
The National Key Research and Development Program of China(2019YFE0111600);The National Natural Science Foundation of China(61971083);The National Natural Science Foundation of China(51939001);The Dalian Science and Technology Innovation Fund(2019J11CY015)

摘要/Abstract

摘要：

广义频分复用（GFDM）系统存在固有的子载波间干扰和子符号间干扰，在时频双选择信道下，会产生严重的导频污染现象，使基于导频的信道估计性能显著下降。为此，提出一种基于深度学习的 GFDM 系统信道估计框架，将离散导频位置处最小二乘信道估计值构成低分辨图像作为网络输入，利用深度残差网络恢复信道时频响应的高分辨图像，实现GFDM系统的信道估计。设计了基于深度残差网络的GFDM时频双选择信道估计算法的仿真系统，通过离线训练获得深度残差网络的最优参数。仿真结果表明，所提算法能够得到接近于最小均方误差信道估计的精度和误码率性能，并具有稳健的多普勒频移泛化能力。

关键词: 广义频分复用, 深度学习, 双选择信道, 信道估计, 残差网络

Abstract:

There exist intrinsic inter-carrier interference and inter-subsymbol interference in generalized frequency division multiplexing (GFDM) systems.Under condition of time-frequency doubly selective channels, severe effects of pilot contamination would occur and lead to significant performance degradation for the pilot-based channel estimations.To this end, a channel estimation framework for GFDM systems based on deep learning was proposed, which took the low-resolution image constructed with the least squares estimated channel gains of the pilot symbols as input.Consequently, a high-resolution image about the channel time-frequency response was recovered through a deep residual network, and the channel estimation was achieved for GFDM systems.A simulation system for the proposed GFDM time-frequency doubly selective channel estimation algorithm based on deep residual network was developed, and the optimal parameters of the deep residual network were determined through an offline training process.Simulation results show that the proposed algorithm can achieve better performance near to minimum mean square error (MMSE) estimation in terms of estimation error and bit error rate (BER), and has robust Doppler frequency shift generalization capability.

Key words: generalized frequency division multiplexing (GFDM), deep learning, doubly-selective channel, channel esti-mation, residual network

中图分类号:

TN92

王莹, 任军, 史可, 林彬. 基于深度学习的广义频分复用系统时频双选择信道估计[J]. 通信学报, 2021, 42(10): 233-242.

Ying WANG, Jun REN, Ke SHI, Bin LIN. Doubly-selective channel estimation for generalized frequency division multiplexing systems based on deep learning[J]. Journal on Communications, 2021, 42(10): 233-242.

图/表 14

图1

图2

图3

图4

表1

图5

图6

图7

表2

3种信道估计算法的计算复杂度"

方法	参数数量	实数乘法次数
CE-DNN	$\sum_{l = 1}^{L - 1} (N_{l} + 1) N_{l + 1}$	$\sum_{l = 1}^{L - 1} N_{l} N_{l + 1}$
MMSE	$K M \| P \| + \| P \|^{2}$	$4 K M (\| P \|^{2} + \| P \|)$
CE-ResNet	$\sum_{l = 1}^{L - 1} (k_{l, h} k_{l, w} k_{l, c} + 1) C_{l + 1}$	$\sum_{l = 1}^{L - 1} k_{l, h} k_{l, w} k_{l, c} H_{l + 1} W_{l + 1} C_{l + 1}$

表2

表3

图8

图9

图10

图11

参考文献 18

[1]	SHARMA S K , WANG X B . Toward massive machine type communications in ultra-dense cellular IoT networks:current issues and machine learning-assisted solutions[J]. IEEE Communications Surveys ＆Tutorials, 2020,22(1): 426-471.
[2]	YANG H J , ZHANG K , ZHENG K ,et al. Joint frame design and resource allocation for ultra-reliable and low-latency vehicular networks[J]. IEEE Transactions on Wireless Communications, 2020,19(5): 3607-3622.
[3]	吴虹, 刘兵, 赵迎新 ,等. 广义频分复用通信系统抑制干扰技术[J]. 上海交通大学学报, 2017,51(9): 1117-1123.
	WU H , LIU B , ZHAO Y X ,et al. Research on interference suppression in generalized frequency division multiplex system[J]. Journal of Shanghai Jiao Tong University, 2017,51(9): 1117-1123.
[4]	黄翔东, 王惠杰, 黎鸣诗 ,等. GFDM 系统低复杂度最小均方误差接收机解调算法[J]. 北京邮电大学学报, 2019,42(3): 7-13.
	HUANG X D , WANG H J , LI M S ,et al. Low-complexity MMSE demodulation algorithm for GFDM[J]. Journal of Beijing University of Posts and Telecommunications, 2019,42(3): 7-13.
[5]	MICHAILOW N , MATTHé M , GASPAR I S ,et al. Generalized frequency division multiplexing for 5th generation cellular networks[J]. IEEE Transactions on Communications, 2014,62(9): 3045-3061.
[6]	VILAIPORNSAWAI U , JIA M . Scattered-pilot channel estimation for GFDM[C]// Proceedings of 2014 IEEE Wireless Communications and Networking Conference (WCNC). Piscataway:IEEE Press, 2014: 1053-1058.
[7]	EHSANFAR S , MATTHE M , ZHANG D ,et al. Interference-free pilots insertion for MIMO-GFDM channel estimation[C]// Proceedings of 2017 IEEE Wireless Communications and Networking Conference (WCNC). Piscataway:IEEE Press, 2017: 1-6.
[8]	EHSANFAR S , MATTHE M , ZHANG D ,et al. Theoretical analysis and CRLB evaluation for pilot-aided channel estimation in GFDM[C]// Proceedings of 2016 IEEE Global Communications Conference (GLOBECOM). Piscataway:IEEE Press, 2016: 1-7.
[9]	LI F , ZHENG K , ZHAO L ,et al. Design and performance of a novel interference-free GFDM transceiver with dual filter[J]. IEEE Transactions on Vehicular Technology, 2019,68(5): 4695-4706.
[10]	JEONG J , PARK Y , WEON S ,et al. Eigendecomposition-based GFDM for interference-free data transmission and pilot insertion for channel estimation[J]. IEEE Transactions on Wireless Communications, 2018,17(10): 6931-6943.
[11]	桂冠, 王禹, 黄浩 . 基于深度学习的物理层无线通信技术:机遇与挑战[J]. 通信学报, 2019,40(2): 19-23.
	GUI G , WANG Y , HUANG H . Deep learning based physical layer wireless communication techniques:opportunities and challenges[J]. Journal on Communications, 2019,40(2): 19-23.
[12]	梁应敞, 谭俊杰, NIYATO D . 智能无线通信技术研究概况[J]. 通信学报, 2020,41(7): 1-17.
	LIANG Y C , TAN J J , NIYATO D . Overview on intelligent wireless communication technology[J]. Journal on Communications, 2020,41(7): 1-17.
[13]	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.
[14]	SOLTANI M , POURAHMADI V , MIRZAEI A ,et al. Deep learning-based channel estimation[J]. IEEE Communications Letters, 2019,23(4): 652-655.
[15]	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.
[16]	HE K M , ZHANG X Y , REN S Q ,et al. Deep residual learning for image recognition[C]// Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2016: 770-778.
[17]	HE K M , SUN J . Convolutional neural networks at constrained time cost[C]// Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2015: 5353-5360.
[18]	ZEILER M D , KRISHNAN D , TAYLOR G W ,et al. Deconvolutional networks[C]// Proceedings of 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2010: 2528-2535.

参数名称	参数取值
系统带宽/MHz	10
子载波数	K=64
子符号数	M=13
导频符号数目	K_p×M_p=22×4
原型滤波器	根升余弦函数
滚降因子α	0.1
调制方式	4-QAM
时变信道模型	Jakes
多径时延/ns	[0, 300, 500, 600, 800]
路径功率/dB	[0, -8, -17, -21, -25]

基于深度学习的广义频分复用系统时频双选择信道估计

Doubly-selective channel estimation for generalized frequency division multiplexing systems based on deep learning

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 14

参考文献 18

相关文章 15

Metrics

推荐阅读 0

[1]	陈东昱, 陈华, 范丽敏, 付一方, 王舰. 基于深度学习的随机性检验策略研究[J]. 通信学报, 2023, 44(6): 23-33.
[2]	李荣鹏, 汪丙炎, 张宏纲, 赵志峰. 知识增强的语义通信接收端设计[J]. 通信学报, 2023, 44(6): 70-76.
[3]	马帅, 裴科, 祁华艳, 李航, 曹雯, 王洪梅, 熊海良, 李世银. 基于生成模型的地磁室内高精度定位算法研究[J]. 通信学报, 2023, 44(6): 211-222.
[4]	崔伟, 于颖, 于海霞, 陈超, 李云鹏. 基于IOC-CSMP的OFDM系统稀疏信道快速重构算法[J]. 通信学报, 2023, 44(2): 52-58.
[5]	曾嵘, 杭潇. 车联网环境下可重构智能反射面辅助无线信道估计算法[J]. 通信学报, 2022, 43(8): 142-150.
[6]	杨洁, 董标, 付雪, 王禹, 桂冠. 基于轻量化分布式学习的自动调制分类方法[J]. 通信学报, 2022, 43(7): 134-142.
[7]	杨秀璋, 彭国军, 李子川, 吕杨琦, 刘思德, 李晨光. 基于Bert和BiLSTM-CRF的APT攻击实体识别及对齐研究[J]. 通信学报, 2022, 43(6): 58-70.
[8]	廖勇, 王世义. 高速移动环境下基于RM-Net的大规模MIMO CSI反馈算法[J]. 通信学报, 2022, 43(5): 166-176.
[9]	廖育荣, 王海宁, 林存宝, 李阳, 方宇强, 倪淑燕. 基于深度学习的光学遥感图像目标检测研究进展[J]. 通信学报, 2022, 43(5): 190-203.
[10]	赵增华, 童跃凡, 崔佳洋. 基于域自适应的Wi-Fi指纹设备无关室内定位模型[J]. 通信学报, 2022, 43(4): 143-153.
[11]	蒋占军, 刘欢, 张华卫, 李翠然. OTFS接收端信道估计辅助的无记忆功放非线性失真补偿算法[J]. 通信学报, 2022, 43(12): 134-145.
[12]	邢旺, 唐晓刚, 周一青, 张冲, 潘振岗. 面向OTFS的时延-多普勒域信道估计方法综述[J]. 通信学报, 2022, 43(12): 188-201.
[13]	廖勇, 程港, 李玉杰. 基于深度展开的大规模MIMO系统CSI反馈算法[J]. 通信学报, 2022, 43(12): 77-88.
[14]	段雪源, 付钰, 王坤, 李彬. 基于简单统计特征的LDoS攻击检测方法[J]. 通信学报, 2022, 43(11): 53-64.
[15]	霍俊彦, 邱瑞鹏, 马彦卓, 杨付正. 基于最邻近帧质量增强的视频编码参考帧列表优化算法[J]. 通信学报, 2022, 43(11): 136-147.

方法	参数数量	实数乘法次数
MMSE	80 960	26 064 896
CE-DNN	8 774 164	8 768 000
CE-ResNet	53 298	4 883 200