tLoc: A Mathematical Probabilistic Model for Wireless Fingerprinting

Avstract- Wireless local area network (WLAN) fingerprint-based positioning is a method to locate devices in indoor environments, in which established solutions such as GPS are unusable or unreliable. It consists in using measurements of Received Signal Strength (RSS) from Access Points to build a radio map of the area, and localize devices by comparing actual RSS measurements to the radio map. This work proposes a novel probabilistic model to estimate precise indoor positioning. To our best knowledge, this is the first work to use such model with a classical likelihood estimation approach for the indoor localization problem. The validation of the model was done using UJIIndorloc, a public database recorded on Universitat Jaume I, in which the tLoc approched to achieve a mean error of 6.94m, while using simple and explainable physical model.