摘要
针对传统室内定位方法在准确性及稳定性上的不足,本文提出了一种基于信道状态信息(channel state information,CSI)的无源室内定位方法。该方法采用普通设备搭建了实验平台,离线阶段采集CSI数据建立位置指纹库,在线阶段则利用机器学习的朴素贝叶斯算法进行位置分类。为进一步提高分类准确度,本文还提出了置信度方法,通过综合多条天线对的结果来减少位置误判。实验结果表明,本文所提出方法能有效实现对室内人员的无源定位,可以达到90%以上的准确度。
To deal with the shortcomings of indoor localization in terms of accuracy and stability, a method based on channel state information （CSI） was proposed to realize passive indoor positioning. A platform was set up with off- the-shelf equipment to collect CSI data. During the offline stage of the method, gathered data from each location were stored as a fingerprint in the database. During the online stage, naive Bayes classification from machine learning was utilized to classify the locations. Furthermore, the degree of confidence was proposed to combine the esti-mation from different antenna pairs. Result shows that the proposed method can effectively realize passive indoor human localization with an accuracy of more than 90% .
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2017年第8期1328-1334,共7页
Journal of Harbin Engineering University
基金
国家自然科学基金项目(61273212)
浙江省自然科学基金项目(LY13F010011
LY14F050004)
浙江省科技厅重大专项项目(2014NM002)
关键词
无源定位
正交频分复用
多输入多输出
信道状态信息
异常值
指纹库
朴素贝叶斯分类
置信度
passive localization
orthogonal frequency division multiplexing
multiple input multiple output
chan-nel state information
outlier
fingerprint
naive Bayes classification
degree of confidence
