Opportunistically supported ubiquitous localization: Machine learning enhancements

Abstract

A lot of work has already been done on the area of localization of mobile nodes, but there still exist numerous open issues. The constant progress in technology gives the opportunity to obtain efficient, in terms of cost and accuracy, localization services, and likewise it increases the number of challenges to be considered. The main problems related to the localization of mobile devices concern with the heterogeneity of visited environments and interested hardware platforms, the energy and computational constraints imposed by the devices and the choice of appropriate tracking technologies. The most important goal in the localization research area is to provide a very accurate positioning service, regardless the surrounding environment. For my PhD research I propose an innovative system for ubiquitous localization which does not rely on backend servers (each node performs a self-localization) and whose reference platform is a new generation mobile phone. This system exploits the sensors embedded on such devices to perform positioning and it is further assisted by opportunistic exchange of information among neighboring nodes. By means of this system, a node is able to classify its movements by continuously refining a self-movement-model (machine learning techniques), thus assisting the localization procedure itself. The purpose of this document is to briefly describe the state of the art in localization, and to outline my planned PhD research.