TY - JOUR
T1 - Toward Scalable and Robust Indoor Tracking
T2 - Design, Implementation, and Evaluation
AU - Jin, Feiyu
AU - Liu, Kai
AU - Zhang, Hao
AU - NG, Joseph K Y
AU - Guo, Songtao
AU - Lee, Victor C.S.
AU - Son, Sang H.
N1 - Funding Information:
This work was supported in part by the National Natural Science Foundation of China under Grant 61872049 and Grant 61572088, in part by the Frontier Interdisciplinary Research Funds for the Central Universities under Project 2018CDQYJSJ0034, in part by the Hong Kong Baptist University Research Centre for Ubiquitous Computing, in part by the Innovation and Technology Commission of the HK SAR Government under the Innovation and Technology Fund under Project ITP/048/14LP, in part by the Global Research Laboratory Program under through NRF under Grant 2013K1A1A2A02078326, and in part by the ICT Research and development Program of MSIP/IITP (Resilient Cyber-Physical Systems Research) under Grant 2014-0-00065.
PY - 2020/2
Y1 - 2020/2
N2 - Although indoor localization has been studied over a decade, it is still challenging to enable many IoT applications, such as activity tracking and monitoring in smart home and customer navigation and trajectory mining in smart shopping mall, which typically require meter-level localization accuracy in a highly dynamic and large-scale indoor environment. Therefore, this article aims at designing and implementing an adaptive and scalable indoor tracking system in a cost-effective way. First, we propose a zero site-survey overhead (ZSSO) algorithm to enhance the system scalability. It integrates the step information and map constraints to infer user's positions based on the particle filter and supports the auto labeling of scanned Wi-Fi signal for constructing the fingerprint database without the extra site-survey overhead. Further, we propose an iterative-weight-update (IWU) strategy for ZSSO to enhance system robustness and make it more adaptive to the dynamic changing of environments. Specifically, a two-step clustering mechanism is proposed to delete outliers in the fingerprint database and alleviate the mismatch between the auto-tagged coordinates and the corresponding signal features. Then, an iterative fingerprint update mechanism is designed to continuously evaluate the Wi-Fi fingerprint localization results during online tracking, which will further refine the fingerprint database. Finally, we implement the indoor tracking system in real-world environments and conduct a comprehensive performance evaluation. The field testing results conclusively demonstrate the scalability and effectiveness of the proposed algorithms.
AB - Although indoor localization has been studied over a decade, it is still challenging to enable many IoT applications, such as activity tracking and monitoring in smart home and customer navigation and trajectory mining in smart shopping mall, which typically require meter-level localization accuracy in a highly dynamic and large-scale indoor environment. Therefore, this article aims at designing and implementing an adaptive and scalable indoor tracking system in a cost-effective way. First, we propose a zero site-survey overhead (ZSSO) algorithm to enhance the system scalability. It integrates the step information and map constraints to infer user's positions based on the particle filter and supports the auto labeling of scanned Wi-Fi signal for constructing the fingerprint database without the extra site-survey overhead. Further, we propose an iterative-weight-update (IWU) strategy for ZSSO to enhance system robustness and make it more adaptive to the dynamic changing of environments. Specifically, a two-step clustering mechanism is proposed to delete outliers in the fingerprint database and alleviate the mismatch between the auto-tagged coordinates and the corresponding signal features. Then, an iterative fingerprint update mechanism is designed to continuously evaluate the Wi-Fi fingerprint localization results during online tracking, which will further refine the fingerprint database. Finally, we implement the indoor tracking system in real-world environments and conduct a comprehensive performance evaluation. The field testing results conclusively demonstrate the scalability and effectiveness of the proposed algorithms.
KW - Algorithm design
KW - indoor localization
KW - performance evaluation
KW - trajectory tracking
KW - Wi-Fi fingerprint
UR - http://www.scopus.com/inward/record.url?scp=85079739632&partnerID=8YFLogxK
U2 - 10.1109/JIOT.2019.2953376
DO - 10.1109/JIOT.2019.2953376
M3 - Journal article
AN - SCOPUS:85079739632
SN - 2327-4662
VL - 7
SP - 1192
EP - 1204
JO - IEEE Internet of Things Journal
JF - IEEE Internet of Things Journal
IS - 2
M1 - 8897653
ER -