深度学习辅助的5G OFDM系统的信道估计  

Deep Learning Assisted Channel Estimation for 5G OFDM Systems

作　　者：王义元 常俊[1] 卢中奎 余福慧 魏家齐 WANG Yiyuan;CHANG Jun;LU Zhongkui;YU Fuhui;WEI Jiaqi(School of Information,Yunnan University,Kunming 650500,China)

机构地区：[1]云南大学信息学院,昆明650500

出　　处：《电讯技术》2024年第1期36-42,共7页Telecommunication Engineering

基　　金：国家自然科学基金资助项目(61562090)。

摘　　要：传统的信道估计算法难以满足5G系统中的高速率低时延的需求。针对该问题,将通信信道的时频响应视为二维图像,提出了一种基于图像恢复技术的信道估计方法。首先,设定参数产生基于5G新空口(New Radio,NR)标准的物理下行链路共享信道(Physical Downlink Shared Channel,PDSCH)的信道数据信息数据集,将所产生的信道矩阵看作二维图像;然后,构建基于卷积神经网络的图像恢复网络,并融入残差连接来提高网络的性能;最后,利用训练好的网络模型进行信道估计。仿真结果表明,与最小二乘算法(Least Square,LS)、实际信道估计(Practical Channel Estimation,PCE)和基于图像超分辨率ChannelNet网络相比,所提出的信道估计算法性能提升明显。The traditional channel estimation algorithm is difficult to meet the requirement of high speed and low delay in 5G system.For this problem,the authors propose a channel estimation method based on image restoration technology by considering the time-frequency response of communication channel as a two-dimensional image.First,parameters are set to generate a channel data information data set of physical downlink shared channel(PDSCH)based on 5G new radio(NR)standard,and the generated channel matrix is treated as a two-dimensional image.Then,an image restoration network based on convolutional neural network is constructed,and residual connection is incorporated to improve the performance of the network.Finally,the trained network model is used for channel estimation.The simulation results show that the performance of the proposed channel estimation algorithm is significantly improved compared with those of the Least Square(LS),Practical Channel Estimation(PCE)and Image-based Super-resolution ChannelNet network.

关 键 词：5G 正交频分复用(OFDM) 信道估计 深度学习 卷积神经网络 

分 类 号：TN911[电子电信—通信与信息系统]