Acquiring accurate vehicle location information in urban settings is very challenging due to the complexity of urban environments. In this paper, we propose a novel scheme, called UPS, to tackle urban vehicle localization problem. After extensive empirical study, we find that GSM power spectrogram collected over a distance has ideal temporal-spatial characteristics for fingerprinting. Encouraged by this observation, UPS tries to utilize the geographical trajectory and the associated GSM power spectrogram information of a moving vehicle to identify its location with reference to a map. To this end, two appealing techniques, i.e., emph{online vehicle localization} and emph{GSM map construction}, are elegantly integrated. With the former, a vehicle can accurately fix its location under complex urban environments. With the latter, a reliable metropolitan-scale GSM power map can be cost-efficiently built at edges, leveraging the strong power of crowdsourcing. By design, UPS is light-weight, requiring only a minimum hardware deployment. We implement a prototype system to validate the feasibility of the UPS design. Furthermore, we conduct extensive trace-driven simulations and results show that UPS can work stably in various urban settings and achieve an accuracy of 5.3 meters with a 90% precision, overwhelming the performance of GPS by five times.

PDF