A Robust WLAN Positioning System Based on Probabilistic Propagation Model

User localization is the key to enable many location-based services. Localization techniques have been actively studied recently due to service as well as safety and security matters. There are many localization systems with different architectures, configurations, accuracies and reliabilities. However, no technique is now generally applicable for indoor environments. In this paper, a probabilistic propagation model of WiFi signals in an environment with walls/floors, and its application to a localization technique using WiFi signal strength, are introduced. A small program installed on the smart device regularly scans the signal strengths to surrounding WiFi access points and send information to a central server to calculate the position. The user location is achieved in 3D space by a robust hybrid method which is based on the probabilistic propagation model and takes the advantages from both well-known geometrical-calculation and finger-printing approaches: low site-survey cost and robustness. The parameters are tuned by using a genetic algorithm from easily collected data.