Estimating and Learning the Trajectory of Mobile Phones
This project is based on the ongoing data collection campaign by Nokia Re- search Center-Lausanne. We use location data sampled everyday by mobile phones in the campaign to estimate position of the participants. It is with emerging mobile systems that combines different sensors in a mobile phone so, we can merge different information sources to improve our estimations. Positioning is a problem encountered frequently in many applications. GPS is widely used for positioning but its output is noisy and it does not work in every location. Considering the embedded sensors and processing capacity of mobile phone, we can improve positioning of clients by other data like visited GSM cell and Wireless LANs. In addition, for situations where GPS does not work such as indoors, we can lay on these information to locate the user. Although most mobile phones are equipped with GPS receivers, users prefer to keep them turned off because of their considerable battery consumption. On the other hand, people usually take determined paths when they want to travel among places that they frequently visit. Learning these trajectories helps to keep GPS receiver turned off and localize users by other sources of information. We describe a new positioning algorithm with two modes of operations: one for cases that we have GPS signals and the other for times that there is not any GPS signal. This work also outlines a new algorithm, based on undergoing Nokia data collection campaign, for positioning and navigating of participants among their recurrent locations.