[Back]

G. Retscher, A. Leb:
"Impact of Different Wi-Fi RSS Scan Durations on Kinematic Smartphone Positioning";
Talk: 11th International Conference on Mobile Mapping Technology 2019, Shenzhen, China (invited); 2019-05-06 - 2019-05-08; in: "The 11th International Conference on Mobile Mapping Technology (MMT 2019)", (2019), Paper ID 26, 8 pages.

Wi-Fi fingerprinting is nowadays a popular method for indoor positioning with smartphones, whereby for system training received signal strengths (RSS) of the surrounding Access Points (APs) are usually measured in static mode at reference points for the time being. From these training observations so-called radio maps are derived with which the current measurements in the online positioning phase can then be compared and the userīs position determined. In this work, static training measurements are completely not foreseen which implies a much greater challenge. The smartphone user does not stop at the required reference points and performs the system training in kinematic mode while walking with normal speed along his trajectory from the start to the desired destination. Experiments using three different smartphones were conducted in an indoor office and transitional outdoor environment to prove the new training concept. In this paper, the results of two trajectories are presented which start in front of the entrances of the building leading through the ground floor and ending in the courtyard of the building. It turned out that the results can vary significantly depending on the smartphone used, which is mainly caused by the duration of a single Wi-Fi scan. This scan duration depends not only on the number of visible APs but mainly on the embedded Wi-Fi chip in the smartphone. In the experiments, it was seen that the respective scan duration can range from only 0.5 to over 4 seconds for the different smartphones. Thus, the position determination in the positioning phase resulted then in deviations from the ground truth of about 2 to 5 m, which is only slightly worse than with static training measurements. The big advantage of the proposed approach, however, is that no static system training phase is required and therefore continuous system training can be performed.

Wi-Fi fingerprinting, kinematic training phase, scan duration dependence, time reduction for training, continuous system training