To improve the accuracy and efficiency of time-varying channels estimation algorithms for millimeter Wave(mmWave)massive Multiple-Input Multiple-Output(MIMO)systems in Internet of Vehicles(IoV)scenarios,the paper prop...To improve the accuracy and efficiency of time-varying channels estimation algorithms for millimeter Wave(mmWave)massive Multiple-Input Multiple-Output(MIMO)systems in Internet of Vehicles(IoV)scenarios,the paper proposes a deep learning(DL)algorithm,Squeeze-and-Excitation Attention Residual Network(SEARNet),which integrates Squeeze-and-Excitation Attention(SEAttention)mechanism and residual module.Specifically,SEARNet considers the channel information as an image matrix,and embeds a SEAttention module in residual module to construct the SEAttention-Residual block.Through a data-driven approach,SEARNet can effectively extract key information from the channel matrix using the SEAttention mechanism,thereby reducing noise interference and estimating the channel in an accurate and efficient manner.The simulation results show that compared to two traditional and two DL channel estimation algorithms,the proposed SEARNet can achieve a maximum reduction in normalized mean square error(NMSE)of 97.66%and 84.49%at SNR of-10 dB,78.18%at SNR of 5 dB,and 43.51%at SNR of 10 dB,respectively.展开更多
Cell-free(CF)multiple-input multiple-output(MIMO)is a promising technique to enable the vision of ubiquitous wireless connectivity for next-generation network communications.Compared to traditional co-located massive ...Cell-free(CF)multiple-input multiple-output(MIMO)is a promising technique to enable the vision of ubiquitous wireless connectivity for next-generation network communications.Compared to traditional co-located massive MIMO,CF MIMO allows geographically distributed access points(APs)to serve all users on the same time-frequency resource with spatial multiplexing techniques,resulting in better performance in terms of both spectral efficiency and coverage enhancement.However,the performance gain is achieved at the expense of deploying more APs with high cost and power consumption.To address this issue,the recently proposed reconfigurable intelligent surface(RIS)technique stands out with its unique advantages of low cost,low energy consumption and programmability.In this paper,we provide an overview of RIS-assisted CF MIMO and its interaction with advanced optimization designs and novel applications.Particularly,recent studies on typical performance metrics such as energy efficiency(EE)and spectral efficiency(SE)are surveyed.Besides,the application of RIS-assisted CF MIMO techniques in various future communication systems is also envisioned.Additionally,we briefly discuss the technical challenges and open problems for this area to inspire research direction and fully exploit its potential in meeting the demands of future wireless communication systems.展开更多
针对车联网场景下多入多出-正交时频空(Multiple-Input Multiple-Output-Orthogonal Time Frequency Space,MIMO-OTFS)系统的信道状态信息(Channel State Information,CSI)反馈问题,提出了一种面向时延-多普勒(Delay-Dopler,DD)域CSI反...针对车联网场景下多入多出-正交时频空(Multiple-Input Multiple-Output-Orthogonal Time Frequency Space,MIMO-OTFS)系统的信道状态信息(Channel State Information,CSI)反馈问题,提出了一种面向时延-多普勒(Delay-Dopler,DD)域CSI反馈的时间差分架构Transformer网络(Time-differencing Architecture Delay-Doppler Transformer Network,TA-DD-TransNet),引入分时反馈机制,将残差信息建模与压缩反馈相结合。网络结构融合Transformer的全局建模能力与卷积神经网络的局部特征提取优势,在保持CSI重构精度的同时显著降低了反馈比特数与计算复杂度。在不同车速、信噪比及非完美信道估计条件下的仿真实验结果表明,所提方法在归一化均方误差(Normalized Mean Squared Error,NMSE)和余弦相似度指标上均优于CsiNet、CsiNet+和BCsiNet。在60 km/h、30 dB信噪比、1/4压缩率下,TA-DD-TransNet的NMSE约-27 dB,余弦相似度达0.96。复杂度分析显示,TA-DD-TransNet在1/4压缩率下的编码器和解码器浮点运算次数分别为1.809×10^(7)和2.281×10^(7),参数量均为8.4×10~6左右,显著低于CsiNet+。所提方法能满足车联网中对高可靠低时延通信的实际需求。展开更多
基金supported in part by the National Natural Science Foundation of China under Grants U2001213 and 62261024in part by National Key Research and Development Project under Grant 2020YFB1807204in part by Key Laboratory of Universal Wireless Communications(BUPT),Ministry of Education under Grant KFKT2022101.
文摘To improve the accuracy and efficiency of time-varying channels estimation algorithms for millimeter Wave(mmWave)massive Multiple-Input Multiple-Output(MIMO)systems in Internet of Vehicles(IoV)scenarios,the paper proposes a deep learning(DL)algorithm,Squeeze-and-Excitation Attention Residual Network(SEARNet),which integrates Squeeze-and-Excitation Attention(SEAttention)mechanism and residual module.Specifically,SEARNet considers the channel information as an image matrix,and embeds a SEAttention module in residual module to construct the SEAttention-Residual block.Through a data-driven approach,SEARNet can effectively extract key information from the channel matrix using the SEAttention mechanism,thereby reducing noise interference and estimating the channel in an accurate and efficient manner.The simulation results show that compared to two traditional and two DL channel estimation algorithms,the proposed SEARNet can achieve a maximum reduction in normalized mean square error(NMSE)of 97.66%and 84.49%at SNR of-10 dB,78.18%at SNR of 5 dB,and 43.51%at SNR of 10 dB,respectively.
基金supported in part by ZTE Industry-University-Institute Co⁃operation Funds.
文摘Cell-free(CF)multiple-input multiple-output(MIMO)is a promising technique to enable the vision of ubiquitous wireless connectivity for next-generation network communications.Compared to traditional co-located massive MIMO,CF MIMO allows geographically distributed access points(APs)to serve all users on the same time-frequency resource with spatial multiplexing techniques,resulting in better performance in terms of both spectral efficiency and coverage enhancement.However,the performance gain is achieved at the expense of deploying more APs with high cost and power consumption.To address this issue,the recently proposed reconfigurable intelligent surface(RIS)technique stands out with its unique advantages of low cost,low energy consumption and programmability.In this paper,we provide an overview of RIS-assisted CF MIMO and its interaction with advanced optimization designs and novel applications.Particularly,recent studies on typical performance metrics such as energy efficiency(EE)and spectral efficiency(SE)are surveyed.Besides,the application of RIS-assisted CF MIMO techniques in various future communication systems is also envisioned.Additionally,we briefly discuss the technical challenges and open problems for this area to inspire research direction and fully exploit its potential in meeting the demands of future wireless communication systems.
文摘针对车联网场景下多入多出-正交时频空(Multiple-Input Multiple-Output-Orthogonal Time Frequency Space,MIMO-OTFS)系统的信道状态信息(Channel State Information,CSI)反馈问题,提出了一种面向时延-多普勒(Delay-Dopler,DD)域CSI反馈的时间差分架构Transformer网络(Time-differencing Architecture Delay-Doppler Transformer Network,TA-DD-TransNet),引入分时反馈机制,将残差信息建模与压缩反馈相结合。网络结构融合Transformer的全局建模能力与卷积神经网络的局部特征提取优势,在保持CSI重构精度的同时显著降低了反馈比特数与计算复杂度。在不同车速、信噪比及非完美信道估计条件下的仿真实验结果表明,所提方法在归一化均方误差(Normalized Mean Squared Error,NMSE)和余弦相似度指标上均优于CsiNet、CsiNet+和BCsiNet。在60 km/h、30 dB信噪比、1/4压缩率下,TA-DD-TransNet的NMSE约-27 dB,余弦相似度达0.96。复杂度分析显示,TA-DD-TransNet在1/4压缩率下的编码器和解码器浮点运算次数分别为1.809×10^(7)和2.281×10^(7),参数量均为8.4×10~6左右,显著低于CsiNet+。所提方法能满足车联网中对高可靠低时延通信的实际需求。
