Obtaining accurate trajectories from repeated coarse and inaccurate location logs

Context awareness is a key property for enabling context aware services. For a mobile device, the user's location or trajectory is one of the crucial contexts. One common challenge for detecting location or trajectory by mobile devices is to manage the tradeoff between accuracy and power consumption. Typical approaches are (1) controlling the frequency of usage of sensors and (2) sensor fusion technique. The algorithm proposed in this paper takes a different approach to improve the accuracy by merging repeatedly measured coarse and inaccurate location data from cell tower. The experimental result shows that the mean error distance between the detected trajectory and the ground truth is improved from 44m to 10.9m by merging data from 41 days' measurement.