基于Chan-Taylor和优化BP神经网络的5G室内定位算法  

5G indoor location algorithm based on Chan-Taylor and optimized BP neural network

作　　者：李帅辰 武建锋[1,3] LI Shuaichen;WU Jianfeng(National Time Service Center,Chinese Academy of Sciences,Xi’an 710699,China;School of Integrated Circuits,University of Chinese Academy of Sciences,Beijing 100049,China;School of Electronic,Electrical and Communication Engineering,University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院国家授时中心,西安710699 [2]中国科学院大学集成电路学院,北京100049 [3]中国科学院大学电子电气与通信工程学院,北京100049

出　　处：《中国惯性技术学报》2023年第8期806-813,822,共9页Journal of Chinese Inertial Technology

基　　金：装备技术基础科研项目(E054JK1601)。

摘　　要：为提高复杂环境下5G室内定位精度,针对不同应用场景设计了基于Chan-Taylor和优化BP神经网络的5G室内定位算法。当无样本可用时,提出了融合Chan-Taylor算法,使用Chan算法计算出定位值作为Taylor算法初始值进行迭代计算;当有小样本可用时,采用BP神经网络效果更佳;当有大样本可用时,使用遗传算法改进BP神经网络以提高定位精度。在不同场景下对三种算法进行了对比实验,实验结果表明:无样本可用时,Chan-Taylor算法具有更好的鲁棒性和适用性;在45个样本训练情况下,BP定位精度最高,为0.3649 m;在400个样本训练情况下,GA-BP定位精度最高。To improve the accuracy of 5G indoor positioning in complex environments,a 5G indoor positioning algorithm based on Chan-Taylor and optimized BP neural network is designed for different application scenarios.When no samples are available,the fusion Chan-Taylor algorithm is proposed,and the Chan algorithm is used to calculate the localization value as the initial value of the Taylor algorithm for iterative calculation;When a small number of samples are available,BP neural network is more effective;When a large number of samples are available,genetic algorithm is used to improve the BP neural network to improve positioning accuracy.Comparative experiments are conducted on three algorithms in different scenarios,and the experimental results show that the Chan-Taylor algorithm has better robustness and applicability when no samples are available;In the case of 45 samples training,BP has the highest positioning accuracy of 0.3649 m;In the case of 400 samples training,GA-BP has the highest positioning accuracy.

关 键 词：室内定位 5G定位 到达时间差 Chan-Taylor算法 BP神经网络 GA-BP神经网络 

分 类 号：TN929.5[电子电信—通信与信息系统]