Open AccessHILS: hybrid indoor localisation system using Wi‐Fi received signal strength and inertial sensor's measurements of smart‐phone
Author(s) -
Tiwari Sushil,
Jain Vinod kumar
Publication year - 2019
Publication title -
iet communications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.355
H-Index - 62
eISSN - 1751-8636
pISSN - 1751-8628
DOI - 10.1049/iet-com.2018.5845
Subject(s) - trilateration , computer science , real time computing , testbed , benchmark (surveying) , unavailability , inertial measurement unit , dead reckoning , signal (programming language) , mobile phone , reset (finance) , step detection , simulation , embedded system , artificial intelligence , node (physics) , telecommunications , global positioning system , acoustics , computer network , engineering , geodesy , financial economics , reliability engineering , economics , geography , physics , programming language
This study presents a smart‐phone‐based hybrid indoor localisation system (HILS), which integrates two different localisation approaches, i.e., pedestrian dead reckoning (PDR) and Wi‐Fi‐based localisation to compensate the limits of each other. In PDR, the localisation accuracy degrades continuously when the moving distance increases from the starting reference point. However, Wi‐Fi‐based trilateration method requires the Wi‐Fi signals from at least three access points (APs). In this study, the authors propose a more efficient novel Heron‐bilateration‐based position determination (HBPD) technique, which requires the Wi‐Fi signals from only two APs to locate the mobile device. The proposed HILS utilises the HBPD technique to reset the bias drift errors involved with inertial sensors, and compensate the unavailability of strong Wi‐Fi signals from APs using the PDR method. In this work, an experimental testbed is developed to collect the Wi‐Fi signals and evaluate the proposed method. The proposed HILS is also compared with five other benchmark methods for two different trajectories and the experimental results show that the mean localisation error is decreased up to 1.68 and 1.96 m, which indicate the improvement by 16.83 and 20.97% for both the trajectories, respectively, in comparison with the best performing method among the benchmark methods.