Enhanced Wi-Fi Fingerprinting with Building Structure and User Orientation

Wi-Fi fingerprinting has been accepted as a common solution to indoor positioning. However, due to signal fluctuation and external interference, Wi-Fi fingerprinting suffers from aliasing, which may result in degradation of accuracy and being undistinguishable of different floors in a multi-floor environment. This paper proposes two algorithms trying to eliminate or minimize the problems and improve the accuracy and stability of positioning results. The first algorithm models the interior building structure as a connected graph and applies it to the Bayesian positioning algorithm. Experiments in a multi-floor building show that the enhanced algorithm is able to improve the accuracy and stability obviously and reduce wrong estimations of floors effectively. The second algorithm integrates user orientation into the Bayesian positioning algorithm and the experimental results are quite promising. As a conclusion, this paper summarizes the common issues that should be considered when deploying a Wi-Fi fingerprinting system.