基于改进射线跟踪法和BP神经网络算法的室外微蜂窝毫米波信道特性研究  

Millimeter wave channel characteristics of outdoor microcellular based on improved ray tracing method and BP neural network algorithm

作　　者：李双德 刘芫健[1] 林乐科[2] 卞希嘉 朱古月 闫亚欣 任安雯 孙萁 LI Shuangde;LIU Yuanjian;LIN Leke;BIAN Xijia;ZHU Guyue;YAN Yaxin;REN Anwen;SUN Qi(College of Electronic and Optical Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China;China Research Institute of Radio wave Propagation,Qingdao 266107,China)

机构地区：[1]南京邮电大学电子与光学工程学院,南京210023 [2]中国电波传播研究所,青岛266107

出　　处：《电波科学学报》2021年第3期430-442,452,共14页Chinese Journal of Radio Science

基　　金：国家自然科学基金(61871232);江苏省研究生创新计划(KYLX16_0650)。

摘　　要：在28 GHz与39 GHz毫米波频段室外微蜂窝场景下,基于改进射线跟踪法和反向传播(back propagation,BP)神经网络算法对毫米波单发单收信道及单发多收信道进行建模仿真研究.在得到的无线信道仿真数据基础上,研究分析了毫米波信道的路径损耗、均方根(root-mean-square,RMS)时延扩展(delay spread,DS)、接收功率等传播特性.通过与现有文献的测量结果对比分析验证了改进射线跟踪法的正确性与有效性.通过BP神经网络方法拟合得到的路径损耗模型参数结果与改进射线跟踪法仿真得到的路径损耗参数结果对比发现,两者吻合程度很高,验证了BP神经网络算法能很好地对室外微蜂窝毫米波信道大尺度参数进行预测.同时,文中给出了一种普遍适用的用来表征室外微蜂窝视距(line-of-sight,LoS)与非视距(non-line-of-sight,NLoS)场景下28 GHz与39 GHz毫米波信道的路径损耗模型.结果表明:LoS场景下的RMS DS和接收功率都小于NLoS场景下得到的结果;LoS场景与NLoS场景下RMS DS、水平方向到达角、多径簇的个数累积分布函数均服从高斯分布;RMS DS在毫米波频段微蜂窝场景下,随着频率的升高而增大,到达接收端的多径成簇呈现稀疏性.The millimeter-wave single-input single-output(SISO)channel and single input multiple output(SIMO)channel are modeled and simulated in the outdoor microcellular scenarios at 28 GHz and 39 GHz based on improved ray tracing method and back propagation(BP)neural network algorithm.Based on the simulation data of wireless channel,the propagation characteristics of millimeter wave channel,such as path loss,root-mean-square(RMS)delay spread and received power,are analyzed and researched.The correctness and effectiveness of the improved ray tracing method are verified by comparing the measured results of the published literature with the simulated results.The results of path loss model parameters fitted by BP neural network algorithm are compared with those of path loss parameters simulated by improved ray tracing method.It is found that the results of path loss parameters fitted by BP neural network algorithm are in good agreement with those obtained by simulation of improved ray tracing method,which verifies that BP neural network algorithm can well predict large-scale parameters of outdoor microcellular millimeter-wave channel.Meanwhile,a universal path loss model for 28 GHz and 39 GHz millimeter-wave channels in outdoor microcellular line-of-sight(LoS)and non-line-of-sight(NLoS)scenarios is presented.The results show that the RMS delay spread and received power in LoS scenario are smaller than those in NLoS scenario.The cumulative distribution function(CDF)of RMS delay spread,horizontal direction angle of arrival and the number of multipath clusters in LoS and NLoS scenarios are found to fit the Gaussian distribution properly.As the frequency increases,the RMS delay spread tends to increase,and the cluster of multipath arriving at receivers shows sparsity in the microcellular scenarios at millimeter-wave band.

关 键 词：毫米波信道 室外微蜂窝 路径损耗 射线跟踪法 反向传播(BP)神经网络算法 分簇 

分 类 号：TN928[电子电信—通信与信息系统]