Automatic Calibration of Large-scale WiFi-Positioning Systems

The growing market for location-based services causes a great demand for positioning systems with high accuracy. In addition to GPS, there are several WiFi positioning systems which allow a cost-effective indoor and outdoor positioning. But despite the fact that they provide accurate measurements in indoor environments, their outdoor performance is not as good as with GPS. This is due to the fact that outside buildings the access point density is typically lower than indoors and radio signals propagate in a different way. In order to enhance accuracy, modern WiFi positioning systems allow calibrating the radio maps by the use of empirical measurements. But in largescale environments it is nearly impossible to calibrate radio maps manually because of the high expenditure of human labor. In this paper we present an approach for the automatic calibration of WiFi positioning systems assisted by another reference positioning system. By constantly observing all measurements, e.g. on a mobile calibration unit or by a calibration server, it is possible to decide at any moment and place if a calibration is possible and necessary. In that case the calibration process is automatically started and afterwards the accuracy improvements achieved are evaluated. By its modular design the described framework is able to use different reference systems for calibration. Also a combination of multiple systems is possible. A prototype has been implemented and tested at Munich Airport in Germany. The test results prove the applicability of the approach and indicate good accuracy improvements.