Abstract
With the prevailing of mobile phones, more and more location-based services (LBSs) are introduced and they increasingly influence people’s daily life. Take Glympse as an example. It associates the social network with the real world by informing users of the positions of their nearby friends. To obtain positions, generally in these applications, GPS is used. However, the energy consumption of GPS is significant as reported by several previous studies [19] [12]. Many researchers then resort to non-GPS or GPS-less positioning strategies, but at the expenses of positioning precision. Is there a way to keep the power consumption of a positioning system low while maintaining the precision? The key insight of this thesis is that the precision of a non-GPS or less-GPS positioning system can be compensated by a map, which is readily available in today’s smart phones. As long as the position of a user can be marked on the map within the required precision, there will be no need to use any positioning sensor. This entails knowing the starting position, the destination, and the movements of the user. Based on this idea, we propose in this thesis the Destination-aware Non- GPS Positioning System (DNPS). Given a starting point, DNPS applies low-power sensors, such as cell tower id and accelerometers, to position the user on the map with consideration about his/her destination. To figure out the destination of the user, DNPS takes advantage of a Hidden Markov Model to associate the user contexts with his/her daily moving behavior. Our experiments based on real users show that, comparing to GPS and non-GPS/less-GPS systems, DNPS is able to reduce the power consumption while maintain the positioning precision.