针对目前基于信道脉冲响应(Channel Impulse Response,CIR)的非视距(None Line of Sight,NLoS)/视距(Line of Sight,LoS)识别方法精度低、泛化能力差的问题,提出了一种多层卷积神经网络(Convolutional Neural Network,CNN)与通道注意力...针对目前基于信道脉冲响应(Channel Impulse Response,CIR)的非视距(None Line of Sight,NLoS)/视距(Line of Sight,LoS)识别方法精度低、泛化能力差的问题,提出了一种多层卷积神经网络(Convolutional Neural Network,CNN)与通道注意力模块(Channel Attention Module,CAM)相结合的NLoS/LoS识别方法。在多层CNN中嵌入CAM提取原始CIR的时域数据特征,利用全局平均池化层代替全连接层进行特征整合并分类输出。使用欧洲地平线2020计划项目eWINE公开的数据集进行不同结构模型和不同识别方法的对比实验,结果表明,所提出的CNN-CAM模型LoS和NLoS召回率分别达到了92.29%与87.71%,准确率达到了90.00%,F1分数达到了90.22%。与现有多种传统识别方法相比,均具有更好的识别效果。展开更多
Cell-free massive multiple-input multipleoutput(MIMO)is a promising technology for future wireless communications,where a large number of distributed access points(APs)simultaneously serve all users over the same time...Cell-free massive multiple-input multipleoutput(MIMO)is a promising technology for future wireless communications,where a large number of distributed access points(APs)simultaneously serve all users over the same time-frequency resources.Since users and APs may locate close to each other,the line-of-sight(Lo S)transmission occurs more frequently in cell-free massive MIMO systems.Hence,in this paper,we investigate the cell-free massive MIMO system with Lo S and non-line-of-sight(NLo S)transmissions,where APs and users are both distributed according to Poisson point process.Using tools from stochastic geometry,we derive a tight lower bound for the user downlink achievable rate and we further obtain the energy efficiency(EE)by considering the power consumption on downlink payload transmissions and circuitry dissipation.Based on the analysis,the optimal AP density and AP antenna number that maximize the EE are obtained.It is found that compared with the previous work that only considers NLo S transmissions,the actual optimal AP density should be much smaller,and the maximized EE is actually much higher.展开更多
In the time-difference-of-arrival(TDOA)localization,robust least squares(LS)problems solved by mathematical programming were proven to be superior in mitigating the effects of non-line-of-sight(NLOS)propagation.Howeve...In the time-difference-of-arrival(TDOA)localization,robust least squares(LS)problems solved by mathematical programming were proven to be superior in mitigating the effects of non-line-of-sight(NLOS)propagation.However,the existing algorithms still suffer from two disadvantages:1)The algorithms strongly depend on prior information;2)The approaches do not satisfy the mean square error(MSE)optimal criterion of the measurement noise.To tackle the troubles,we first formulate an MSE minimization model for measurement noise by taking the source and the NLOS biases as variables.To obtain stable solutions,we introduce a penalty function to avoid abnormal estimates.We further tackle the nonconvex locating problem with semidefinite relaxation techniques.Finally,we incorporate mixed constraints and variable information to improve the estimation accuracy.Simulations and experiments show that the proposed method achieves consistent performance and good accuracy in dynamic NLOS environments.展开更多
减小非视距(Non Line Of Sight,NLOS)误差定位算法大多要求在移动台和基站之间至少存在一条视距(Line Of Sight,LOS)路径。提出一种新的NLOS环境中基于散射模型分类传播环境的TOA(Time Of Arrival)定位方法,将散射模型中NLOS传播的统计...减小非视距(Non Line Of Sight,NLOS)误差定位算法大多要求在移动台和基站之间至少存在一条视距(Line Of Sight,LOS)路径。提出一种新的NLOS环境中基于散射模型分类传播环境的TOA(Time Of Arrival)定位方法,将散射模型中NLOS传播的统计特性加入到定位算法中,使用散射模型研究了3种定位算法,方差匹配算法,期望最大算法和贝叶斯算法。并对算法进行仿真,仿真结果表明,本算法性能优于传统定位算法。展开更多
提出了三种改进的用卡尔曼滤波器消除到达时间(Time of Arrival, TOA)测量值中非视距(Non-Line ofSight, NLOS)误差的方法。这三种方法从不同角度考察 TOA 测量值中 NLOS 误差的特点,分别对卡尔曼滤波器的迭代过程进行改进,有效地消除了...提出了三种改进的用卡尔曼滤波器消除到达时间(Time of Arrival, TOA)测量值中非视距(Non-Line ofSight, NLOS)误差的方法。这三种方法从不同角度考察 TOA 测量值中 NLOS 误差的特点,分别对卡尔曼滤波器的迭代过程进行改进,有效地消除了 TOA 测量值中 NLOS 误差的随机性和正向偏差。与传统的 NLOS 误差消除算法相比,这三种方法均可获得较小的估计误差,并可实现实时处理。展开更多
文摘针对目前基于信道脉冲响应(Channel Impulse Response,CIR)的非视距(None Line of Sight,NLoS)/视距(Line of Sight,LoS)识别方法精度低、泛化能力差的问题,提出了一种多层卷积神经网络(Convolutional Neural Network,CNN)与通道注意力模块(Channel Attention Module,CAM)相结合的NLoS/LoS识别方法。在多层CNN中嵌入CAM提取原始CIR的时域数据特征,利用全局平均池化层代替全连接层进行特征整合并分类输出。使用欧洲地平线2020计划项目eWINE公开的数据集进行不同结构模型和不同识别方法的对比实验,结果表明,所提出的CNN-CAM模型LoS和NLoS召回率分别达到了92.29%与87.71%,准确率达到了90.00%,F1分数达到了90.22%。与现有多种传统识别方法相比,均具有更好的识别效果。
基金supported in part by the National Natural Science Foundation of China under Grant 62171231in part by the Jiangsu Provincial Key Research and Development Program(No.BE2020084-1)。
文摘Cell-free massive multiple-input multipleoutput(MIMO)is a promising technology for future wireless communications,where a large number of distributed access points(APs)simultaneously serve all users over the same time-frequency resources.Since users and APs may locate close to each other,the line-of-sight(Lo S)transmission occurs more frequently in cell-free massive MIMO systems.Hence,in this paper,we investigate the cell-free massive MIMO system with Lo S and non-line-of-sight(NLo S)transmissions,where APs and users are both distributed according to Poisson point process.Using tools from stochastic geometry,we derive a tight lower bound for the user downlink achievable rate and we further obtain the energy efficiency(EE)by considering the power consumption on downlink payload transmissions and circuitry dissipation.Based on the analysis,the optimal AP density and AP antenna number that maximize the EE are obtained.It is found that compared with the previous work that only considers NLo S transmissions,the actual optimal AP density should be much smaller,and the maximized EE is actually much higher.
基金supported by the National Natural Science Foundation of China under Grant No.62101370。
文摘In the time-difference-of-arrival(TDOA)localization,robust least squares(LS)problems solved by mathematical programming were proven to be superior in mitigating the effects of non-line-of-sight(NLOS)propagation.However,the existing algorithms still suffer from two disadvantages:1)The algorithms strongly depend on prior information;2)The approaches do not satisfy the mean square error(MSE)optimal criterion of the measurement noise.To tackle the troubles,we first formulate an MSE minimization model for measurement noise by taking the source and the NLOS biases as variables.To obtain stable solutions,we introduce a penalty function to avoid abnormal estimates.We further tackle the nonconvex locating problem with semidefinite relaxation techniques.Finally,we incorporate mixed constraints and variable information to improve the estimation accuracy.Simulations and experiments show that the proposed method achieves consistent performance and good accuracy in dynamic NLOS environments.
文摘减小非视距(Non Line Of Sight,NLOS)误差定位算法大多要求在移动台和基站之间至少存在一条视距(Line Of Sight,LOS)路径。提出一种新的NLOS环境中基于散射模型分类传播环境的TOA(Time Of Arrival)定位方法,将散射模型中NLOS传播的统计特性加入到定位算法中,使用散射模型研究了3种定位算法,方差匹配算法,期望最大算法和贝叶斯算法。并对算法进行仿真,仿真结果表明,本算法性能优于传统定位算法。
