Locating User Equipments and Access Points using RSSI Fingerprints: A Gaussian Process Approach

Location fingerprinting (LF) is an attractive localization technique which relies on existing infrastructures. The major drawback of LF is the requirement of having an updated fingerprint database. Gaussian Process (GP) is a nonparametric modeling technique which can be used to model the received signal strength indicator (RSSI) and create the fingerprint database based on few training data. GP, at its natural form, is a non-parametric zero-mean process. A mean function in the form of a parametric pathloss model can be used to model the negative offset of RSSI. However, this typically requires the knowledge of the location of the access points (APs). In this work, we combine the non-parametric GP with the parametric pathloss model to model the RSSI. In particular, the location of the APs is optimized jointly with the hyperparameters of the GP in the offline phase and then used in the mean function to locate the mobile user equipments (UEs). Experimental results show that our proposed scheme significantly reduces the calibration efforts and cost while performs as well as tradtional schemes with the whole area fingerprinted.