Towards Indoor Localization Guided by Machine Learning

Abstract

Indoor geolocation has become a major focus in various fields, such as indoor navigation, warehouse logistics, and targeted marketing in shopping centers. Traditional geolocation technologies, like GPS, show their limitations in indoor environments due to signal loss caused by building structures. As a result, Wi-Fi networks and smartphone sensors present a promising alternative for indoor geolocation. This master's thesis aims to analyze existing work and propose a precise and robust indoor geolocation solution by integrating signals from surrounding Wi-Fi networks and smartphone sensors (such as accelerometers, gyroscopes, magnetometers, etc.). The primary objective is to design a data fusion algorithm that effectively combines information from these different sources to achieve accurate real-time localization.