Performance comparison of a probabilistic fingerprint-based indoor positioning system over different smartphones

In this paper a performance comparison of a probabilistic Gaussian-Kernel fingerprint-based indoor positioning method over different smartphones, is presented. The work aims at highlighting the positioning accuracy, the robustness and the consistency of the method by testing it over two different smartphone platforms (i.e., Nokia N95 and Samsung Galaxy S II), within a given area. In more detail, three different variants of the probabilistic approach have been tested: Nearest Neighbor (NN), K-Nearest Neighbor (K-NN) and K Weighted-Nearest Neighbor (KW-NN). Numerical experiments, carried out in an area of around 80 [m²], have shown that the probabilistic fingerprint provides good position accuracy (less than 1.20 [m] of error) for both devices and also robustness when the signal strength acquisitions are reduced. Finally, the similarity of results provided by the two smartphones leads to assert that the probabilistic approach is also consistent with respect to the device employed in the experiments.