Weighted Fingerprinting Localization based on a Multichannel WiFi Map

WiFi localization based on fingerprinting method became popular during last decade. However, collecting information to construct a WiFi signal map is challenging due to high cost. We assume that it is possible to gather such information with cheap cost using crowd sourcing with smartphones. Each user’s current location is estimated by either GPS, WiFi, radio cell signals or manually set by the user and the WiFi scan information is collected into a server. In this paper, we focus on outdoor geolocalization based multichannel WiFi Map where each channel conveys different accuracy of the reference position. The proposed approach optimizes importance weight on the different channels of WiFi map using Least Squares. Gradient Descent algorithm optimizes weight parameter to minimizes localization error on the training set. The experiment shows the proposed approach outperforms localization with equal importance.