[Back]

G. Retscher, J. Joksch:
"Analysis of Nine Vector Distances for Fingerprinting in Multiple-SSID Wi-Fi Networks";
Poster: 7th International Conference Indoor Positioning and Indoor Navigation IPIN 2016, Alcalá de Henares (invited); 2016-10-04 - 2016-10-06; in: "7th International Conference Indoor Positioning and Indoor Navigation IPIN 2016", (2016), ISBN: 978-1-5090-2424-7; Paper ID 190, 5 pages.

In this analytical study, the use of nine different vector distances, i.e., the Manhattan, Euclidean, Chebyshev, Canberra, Cosine, Sorensen, Hellinger, Chi-square and Jeffrey distance, for positioning using Wi-Fi location fingerprinting is evaluated. A testbed in an office environment with a regular grid of reference points was built therefore. RSS measurements to six access points of four different multiple-SSID (Service Set Identifier) Wi-Fi networks are compared. The achievable positioning accuracies depended mainly on the selection of the vector distance and matching algorithm, i.e., the nearest neighbour NN, k-nearest KNN or k-weighted nearest neighbour KWNN approach. In most tests a similar performance was achieved with the Cosine, Euclidean, Hellinger and Chi-square vector distance in combination with the KNN or KWNN. In average the mean distance error (MDE) resulted in 1.4 m if using a single network in multiple-SSID configuration.

Wi-Fi positioning; multiple-SSID networks; location fingerprinting; matching approaches; vector distances; mean distance error