CTS

GPS has been widely used for locating mobile devices on the road map. Due to its high power consumption and poor signal penetration, GPS is unfortunately unsuitable to be used for continuously tracking low-power devices. Compared with GPS-based positioning, cellular-infrastructure-based positioning consumes much less energy, and works in any place covered by the cellular networks. However, the challenges of cellular positioning come from the relatively low accuracy and sampling rate. In this paper, we propose a novel cellular-based trajectory tracking system, namely CTS. It achieves GPS-level accuracy by combining trilateration-based cellular positioning, stationary state detection, and Hidden-Markov-Model-based path recovery. In particular, CTS utilizes basic characteristics of cellular sectors to produce more credible inferences for device locations. To evaluate the performance of CTS, we collaborated with a mobile operator and deployed the system the city of Urumchi, Xinjiang Province of China. We collected the location data of 489,032 anonymous mobile subscribers from cellular networks during 24 hours, and retrieved 201 corresponding GPS trajectories. Our experimental results show that CTS achieves GPS-level accuracy in 95.7% of cases, which significantly outperforms the state-of-the-art solutions.