Gaussian process for RSS-based localisation

Localisation making use of Wi-Fi Signal Strength is a well established research strand, where various approaches have been proposed by the community. In our work we focus on the problem of localizing a device by considering the Received Signal Strength (RSS) of signals the device receives from the wireless access points around it. We propose a two-parts localization method that makes use of the Gaussian Process technique to interpolate signal vectors received during a training phase in order to estimate the unknown position of the device during test. The proposed method has been tested on four publicly available databases proposed by other authors. The comparison with their results shows the superior performance of our novel approach.