A Crowdsourcing Infrastructure for Fingerprint based Indoor Localisation (Diplomarbeit)
Research on solving the problem of indoor localisation has been highly active for over a decade, but current solutions require a tedious map generation step which demands expert knowledge from the user and effort prior to using the technique. The new ideas developed, implemented, and tested in this thesis allow the elimination of the initial data collection step by combining multiple sensor data (sensor fusion) and exchanging the results with other clients (crowd sourcing). A wireless fingerprinting mechanism and a basic client-server infrastructure were implemented on a mobile phone to evaluate this approach. The evaluation of the prototype implementation shows clearly that sensor fusion and crowd based data exchange is the way to go to make indoor localisation a smooth user experience. Still, the option for semi-supervision by the user is important and should not be neglected. Big companies like Google, Microsoft and Nokia have already started to deploy indoor localisation software but the general lack of indoor map data severely hinders and complicates wide application of this new technique. Using the proposed approaches could greatly speed up the successful introduction of indoor localisation and navigation to every day usage.