基于改进的高斯对数信号指纹相似性度量方法  

An improved logarithmic Gaussian distance for signal fingerprint similarity metric

作　　者：李巍 王鸥 于亮亮 同东辉 曹智 LI Wei;WANG Ou;YU Liang-liang;TONG Dong-hui;CAO Zhi(Information and Communication Branch,State Grid Liaoning Electric Power Co.,Ltd.,Shenyang 110006,China)

机构地区：[1]国网辽宁省电力有限公司信息通信分公司

出　　处：《电子设计工程》2019年第23期78-82,共5页Electronic Design Engineering

摘　　要：由于WiFi信号在许多室内环境中普遍存在,因此利用WiFi信号进行室内定位,相比于其他的室内定位方式,如蓝牙定位、超宽带定位、RFID定位等,有着成本低,可推广等优点,是当前研究的热点。WiFi定位中,通常分为离线训练阶段和在线定位阶段。离线训练阶段需要采集室内环境中不同位置的信号指纹生成指纹地图。在线定位阶段中,通过一定的信号指纹距离定义寻找当前最新采集到的信号指纹与指纹地图中最近的k指纹,再通过加权最近邻法(Weighted K Nearest Neighbor,WKNN)求得当前位置估计。文本研究了在线定位中的不同的信号指纹信号距离定义,并且提出一种改进的高斯对数距离。传统的高斯对数距离仅仅利用了信号强度信息(Received Signal Strength,RSS),而本文提出的改进距离,既利用了RSS信息,也利用了信号指纹不同WiFi接入点(Acess Point,AP)的可见性信息。本文的实验部分中,对比了传统的高斯对数距离、传统的基于秩(rank based)的距离与本文提出的距离在实际定位中的精度,证明了本文提出的距离对信号指纹定位的精度有一定提高。Because the WiFi signals are commonly seen in the indoor environment,if it is suitable for indoor positioning.Compared with other indoor positioning methods,such as Bluetooth based,Ultra Wide Band(UWB)based Radio Frequency Identification(RFID)based methods,WiFi based methods have the advantages of low cost and easy-to-promote.Two steps are needed in WiFi based positioning:the offline training step and the online positioning step.During the offline training step,signal fingerprints at different indoor locations should be collected to form a radio map.During the online positioning phase,recently collected fingerprints are compared with those in the radio map and the Weighted K Nearest Neighbor(WKNN)method is adopted to solve for the position estimation.In this paper,we have proposed a new metric for fingerprint matching during the online step.The proposed metric is an improvement on the Logarithmic Gaussian Distance(LDG)metric.The metric has adopted information of both Access Points(AP)availability and Received Signal Strength(RSS).During the experiment,the proposed metric is compared with the traditional rank based metric and LDG based method in terms of positioning accuracy.The results have shown that the proposed metric has improved the positioning accuracy.

关 键 词：WiFi定位 智能手机 高斯对数距离 室内定位 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]