Destination-aware non-GPS positioning using predictive contexts

Location-based Services (LBS) fairly attract people's attention in recent years. In those applications, location is considered to be one of the best indicators for a mobile user. Traditional positioning mechanisms use location sensors to fetch the user location to trigger navigation. However, we propose a new idea to reverse this procedure. We apply navigation to trigger the positioning by the predictive user destination. Once the location sensors can be discarded, the positioning overhead can be significantly reduced. To predict the destination of mobile users, we take advantages of light-weight sensors like an accelerometer, a compass, a digit map and the sequences of cell-tower signal. By constructing a prediction model which is composed of the above user contexts, we can accurately predict the next n-hop destinations and navigate the current position by the predicted routing path. We propose Destination-aware Non-GPS Positioning System (DNPS) to replace traditional GPS in metropolitan, and our experiments show that DNPS can significantly improve battery life by up to 47% with an acceptable precision penalty about 50 meters while compared to existed periodic-GPS mechanism.