Fusing Similarity-Based Sequence and Dead Reckoning for Indoor Positioning Without Training

Abstract

The traditional fingerprinting-based positioning approach usually requires a laborious training phase to collect the measurements in an environment, which is a challenge for applications involving large buildings. In this paper, we propose a novel approach to fuse similarity-based sequence and dead reckoning to track the positions of users in wireless indoor environments. The reference fingerprinting map is constructed without the need for training and is based upon the geometrical relationships of the transmitters, whose positions are known and can be obtained offline. The fingerprint used for online positioning is represented by a ranked sequence of transmitters based on the measured received signal strength (RSS), which is referred to as RSS sequence in this paper. The similarities between this sequence and the reference fingerprints are computed and embedded into a particle filter to locate a user. The displacement estimation from inertial measurement unit is then integrated into the particle filter to track a mobile user. Moreover, the proposed approach can be easily extended to include other sources of sensors. In this paper, frequency modulation radio signal measurements are used as the example to fuse with Wi-Fi measurements to achieve a better tracking accuracy. Extensive experiments are also conducted to evaluate the proposed approach.