Paper Info
Title
Accurate Indoor Localization With Multiple Feature Fusion
Abstract
In recent years, many fingerprint-based localization approaches have been proposed, in which different features (e.g., received signal strength (RSS) and channel state information (CSI)) were used as the fingerprints to distinguish different positions. Although CSI-based approaches usually achieve higher accuracy than RSSI-based approaches, we find that the localization results of different approaches usually compensate with each other, and by fusing different features we can get more accurate localization results than using only single feature. In this paper, we propose a localization method that fusing different features by combining results of different localization approaches to achieve higher accuracy. We first select three most possible candidate positions from all the candidate positions generated by different approaches according to a newly defined metric called confidence degree, and then use the weighted average of them as the position estimation. When there are more than three candidate positions, we use a minimal-triangle principle to break the tie and select three out of them. Our experiments show that the proposed approach achieves median error of 0.5 m and 1.1 m respectively in two typical indoor environments, significantly better than that of approaches using only single feature.
Year
DOI
Venue
2017
10.1007/978-3-319-60033-8_45
WIRELESS ALGORITHMS, SYSTEMS, AND APPLICATIONS, WASA 2017
Keywords
Field
DocType
Indoor localization, Channel state information, Multiple features
Feature fusion,Pattern recognition,Computer science,Fingerprint,Artificial intelligence,Signal strength,RSS,Channel state information,Distributed computing
Conference
Volume
ISSN
Citations 
10251
0302-9743
1
PageRank 
References 
Authors
0.35
16
5
Name
Order
Citations
PageRank
Yalong Xiao181.48
Jianxin Wang22163283.94
Shigeng Zhang353950.80
Haodong Wang410.35
Jiannong Cao55226425.12