Abstract:

For the problem that the single fingerprint location results were large errors and drifting in indoor environment, an improved algorithm of geomagnetic and WiFi fingerprint system was proposed. First, the larger matching area was divided into smaller sub-regions with more distinctive features by K-means clustering method. In the online phase, Wi-Fi fingerprints were used to coarsely locate small areas, and then a further fine-match positioning was performed through the geomagnetic fingerprint positioning system. Experimental results showed that the integration of algorithm reduced the searching area, and greatly reduced the problem of mis-matching. In the experiment, the average positioning error was only 2.17 m, and the maximum positioning error was 3.61 m. Compared with the single fingerprint positioning system, the performance was greatly improved, which proved that the positioning method had certain feasibility and advancement.

Key words: indoor location, WiFi, geomagnetic field, fingerprint, K-means clustering, magnetic matching