A Wi-Fi fingerprinting-based indoor localization approach: M-Weighted position estimation (m-WPE)

Indoor localization is a crucial topic for mobile computing and it has been attracting numerous research groups and labs around the world. Among the many techniques being proposed for indoor localization, fingerprinting based on the use of Wi-Fi signal has attracted continuous attention in academia because of pervasive penetration of wireless LANs (WLANs) and Wi-Fi enabled mobile devices. In the indoor positioning literature, various approaches on the top of k-Nearest Neighbor (k-NN) algorithm have widespread usage in estimating positions. In this paper, we propose and implement a Weighted Position Estimation (WPE) approach based on the most similar k-neighbors and the strongest APs at the validation point. k-NN algorithm is compared with our proposed approach over UJIIndoorLoc indoor positioning database and the performance of k-NN algorithm with different k-values is also evaluated. Experimental results on the UJIIndoorLoc database reveal that our proposed WPE approach provides up to 20% better prediction accuracy for indoor localization compared to the baseline k-NN.