基于PSO-BP神经网络的5G基站位置确定方法  

A method for 5G access point location determination based on PSO-BP neural network

作　　者：杜莹 韦原原 蒲欢欢 DU Ying;WEI Yuanyuan;PU Huanhuan(School of Geography and Tourism,Zhengzhou Normal University,Zhengzhou 450044,China)

机构地区：[1]郑州师范学院地理与旅游学院,郑州450044

出　　处：《测绘工程》2025年第1期47-52,67,共7页Engineering of Surveying and Mapping

基　　金：国家重点研发计划资助项目(2021YFB3900900)。

摘　　要：5G基站位置的确定对室内定位服务和网络安全有着重要意义。首先对5G信道状态信息CSI进行Hample滤波和降维,然后构建基于粒子群优化PSO的误差反向传播BP神经网络信号损耗模型,建立5G CSI和距离的映射关系,最后基于模型预测的距离实现对5G AP的探测。实验采用室外探测室外和室内5G AP的实测数据,结果表明,与BP神经网络相比,基于PSO-BP神经网络的距离预测值更加精确,室外探测室外和室内5G AP的精度分别达到了0.32 m和0.96 m。随着测量方向数的提升,5G AP的定位精度不断提升。当方向数达到5个时,精度提升最为显著。The determination of the location of 5G access points is of great significance for indoor positioning services and network security.This article first performs Hample filtering and dimensionality reduction on the state information of 5G channels;then,a signal loss model based on Particle Swarm Optimization(PSO)for Back Propagation(BP)neural network was constructed,and the mapping relationship between 5G CSI and distance was established;finally,the detection of 5G AP is achieved based on the distance predicted by the model.The experiment used measured data from indoor detection of outdoor and indoor 5G AP,and the results showed that compared with BP neural network,the distance prediction value based on PSO-BP neural network was more accurate.The accuracy of outdoor detection of outdoor and indoor 5G AP reached 0.32m and 0.96m,respectively.As the number of measurement directions increases,the positioning accuracy of 5G base stations continues to improve;when the number of directions reaches 5,the accuracy improvement is most significant.

关 键 词：信道状态信息 AP探测 粒子群优化 BP神经网络 

分 类 号：P228[天文地球—大地测量学与测量工程]