基于轻量级卷积神经网络的CSI图像室内定位  

Lightweight convolutional neural network-based indoor localization of CSI images

作　　者：黄良璜 余敏[1] HUANG Lianghuang;YU Min(College of Computer Information and Engineering,Jiangxi Normal University,Nanchang 330022,China)

机构地区：[1]江西师范大学计算机信息工程学院,南昌330022

出　　处：《全球定位系统》2025年第1期41-47,共7页Gnss World of China

基　　金：中央引导地方科技发展资金跨区域研发合作项目(20222ZDH04090);江西省教育厅研究生创新基金项目(YC2022-s350)。

摘　　要：针对卷积神经网络(convolutional neural network,CNN)计算复杂度高和内存占用大的问题,本文提出了一种基于轻量级CNN的信道状态信息(channel state information,CSI)图像指纹被动定位(LCNNLoc)方法.离线训练阶段,将幅值差矩阵和相位矩阵构造成类似于“RGB”的三通道特征图像;同时设计了一个轻量级CNN架构,利用特征图像作为该框架的输入进行训练,在训练结束时将CNN模型保存为指纹数据库.在线定位阶段,采用概率加权质心方法实现了实时的位置估计.实验结果表明,相较于传统方法,LCNNLoc不仅提升了定位精度,还降低了算法运行耗时.Aiming at the problem of high computational complexity and large memory occupation of convolutional neural network(CNN),this paper proposes a lightweight CNN-based passive localisation method for channel state information(CSI)image fingerprints(LCNNLoc).In the offline training stage,the amplitude difference matrix and phase matrix are constructed into a three-channel feature image similar to"RGB";at the same time,a lightweight CNN architecture is designed,the feature image is used as the input to train the framework,and the CNN model is saved as a fingerprint database at the end of training.In the online positioning stage,real-time position estimation was achieved using a probability weighted centroid method.The experimental results show that compared with the traditional method,LCNNLoc not only improves the positioning accuracy,but also reduces the algorithm running time consuming.

关 键 词：卷积神经网络(CNN) 信道状态信息(CSI) 图像指纹 轻量级网络 概率加权质心方法 

分 类 号：P228[天文地球—大地测量学与测量工程] TN92[天文地球—测绘科学与技术]