WiFi-PDR室内组合定位的无迹卡尔曼滤波算法

针对当前室内定位的应用需求和亟待解决的关键问题,结合城市室内环境下广泛存在的WiFi无线信号以及智能手机传感器信息,提出了一种WiFi无线信号联合行人航迹推算（PDR）的室内定位方法。该方法采用无迹卡尔曼滤波（UKF）算法对WiFi和PDR定位信息进行融合处理,有效克服了WiFi单点定位精度低和PDR存在累计误差的问题。针对融合算法中WiFi指纹匹配计算量大的问题,用k-means聚类算法对WiFi指纹库进行聚类处理,降低了指纹匹配算法的计算量,提高了算法的实时性。通过在华为P6-U06智能手机平台上实际测试,在时间效率上经过聚类处理后系统定位耗时有很大程度的改善,平均降幅为51%,其中最大降...

Full description

Saved in:
Bibliographic Details
Published in测绘学报 Vol. 44; no. 12; pp. 1314 - 1321
Main Author 陈国良 张言哲 汪云甲 孟晓林
Format Journal Article
LanguageChinese
Published 中国矿业大学国土环境与灾害监测国家测绘地理信息局重点实验室,江苏徐州221116%诺丁汉大学,英国诺丁汉NG7 2TU 2015
中国矿业大学环境与测绘学院,江苏徐州221116
Subjects
k-means
WiFi
室内定位
手机传感器
无迹卡尔曼滤波
行人航迹推算
室内定位
UKF
行人航迹推算
无迹卡尔曼滤波
smart phone sensors
Indoor positioning
ped estrian dead reckoning
k-means
手机传感器
WiFi
Online AccessGet full text
ISSN1001-1595
DOI10.11947/j.AGCS.2015.20140691

Cover

More Information
Summary:针对当前室内定位的应用需求和亟待解决的关键问题,结合城市室内环境下广泛存在的WiFi无线信号以及智能手机传感器信息,提出了一种WiFi无线信号联合行人航迹推算（PDR）的室内定位方法。该方法采用无迹卡尔曼滤波（UKF）算法对WiFi和PDR定位信息进行融合处理,有效克服了WiFi单点定位精度低和PDR存在累计误差的问题。针对融合算法中WiFi指纹匹配计算量大的问题,用k-means聚类算法对WiFi指纹库进行聚类处理,降低了指纹匹配算法的计算量,提高了算法的实时性。通过在华为P6-U06智能手机平台上实际测试,在时间效率上经过聚类处理后系统定位耗时有很大程度的改善,平均降幅为51%,其中最大降幅达到64%,最小的也达到了36%;在定位精度上,当室内人员为行走状态时WiFi定位平均误差为7.76m,PDR定位平均误差为4.57m,UKF滤波融合后平均定位误差下降到1.24m。
Bibliography:CHEN GuoLiang;ZHANG Yanzhe;WANG Yunjia;MENG Xiaolin （1. School of Environment Science and Spatial Informotics, Chino University of Mining and Technology, Xuzhou 221116, Chino; 2. Key Laboratory for Land Environment and Disaster Monitoring of SBSM,Chino University of Mining and Technology, Xuzhou 221]16, China; 3. The University of Nottingham, Nottingham NG7 2TU, UK）
11-2089/P
Indoor positioning still faces lots of fundamental technical problems although it has been widely applied.A novel indoor positioning technology by using the smart phone with the assisting of the widely available and economically signals of WiFi is proposed.It also includes the principles and characteristics in indoor positioning.Firstly,improve the system＇s accuracy by fusing the WiFi fingerprinting positioning and PDR（ped estrian dead reckoning）positioning with UKF（unscented Kalman filter）.Secondly,improve the real-time performance by clustering the WiFi fingerprinting with k-means clustering algorithm.An investigation test was conducted
ISSN:1001-1595
DOI:10.11947/j.AGCS.2015.20140691