| 摘要: |
| 蜂窝移动通信环境复杂多变，在基站和移动台之间不可避免会出现电波的非视距（Non-Line-of-Sight,NLOS）传播，使基站和移动台之间的距离测量误差显著增大，导致定位性能急剧下降。为了准确识别出视距（Line-of-Sight,LOS）与非视距传播的基站信号，提出了一种基于随机森林的LOS/NLOS基站识别方法，通过分析移动台与各基站接收机测量距离与定位误差之间的相关性，选择LOS/NLOS测量距离作为特征进行分类器训练，再将分类器用于LOS/NLOS基站的识别。仿真结果表明，该方法对NLOS基站的正确识别率达到90%以上，能取得较好的定位性能。 |
| 关键词: 无线定位 基站信号识别 NLOS识别 机器学习 随机森林 |
| DOI：10.20079/j.issn.1001-893x.220320001 |
|
| 基金项目:国家自然科学基金资助项目(61871332) |
|
| A Random Forest-based LOS/NLOS Base Station Identification Method |
| JI Jinghong,ZHANG Zhenyu,DENG Ping |
| (School of Information Science and Technology,Southwest Jiaotong University,Chengdu 611756,China) |
| Abstract: |
| The cellular mobile communication environment is complex and volatile,and the Non-Line-of-Sight(NLOS) propagation of radio waves inevitably occurs between base stations and mobile stations,which causes a significant increase in the distance measurement error between base stations and mobile stations and leads to a sharp decrease in localization performance.In order to accurately identify Line-of-Sight(LOS) and NLOS base station signals,a Random Forest-based LOS/NLOS base station identification method is proposed.By analyzing the correlations between the measured distances of mobile stations and each base station receiver and the localization errors,the LOS/NLOS measured distances are selected as features for classifier training,and then the classifier is used for the identification of LOS/NLOS base stations.Simulation results show that the correct recognition rate of NLOS base stations by the method reaches more than 90 |
| Key words: wireless positioning base station signal identification NLOS identification machine learning random forest |
