The smartphone as enabler for road traffic information based on cellular network signalling

The higher penetration rate of GPS-enabled smartphones together with their improved processing power and battery life makes them suitable for a number of participatory sensing applications. The purpose of this paper is to analyse how GPS-enabled smartphones can be used in a participatory sensing context to build a radio map for RSS-based positioning, with a special focus on road traffic information based on cellular network signalling. The CEP-67 location accuracy achieved is 75 meters for both GSM and UMTS using Bayesian classification. For this test site, the accuracy is similar for GSM and UMTS, with slightly better results for UMTS in the CEP-95 error metric. The location accuracy achieved is good enough to avoid large errors in travel time estimation for highway environments, especially considering the possibility to filter out estimates with low accuracy using for example the posterior bin probability in Bayesian classification. For urban environments more research is required to determine how the location accuracy will affect the path inference problem in a dense road network. The location accuracy achieved in this paper is also sufficient for other traffic information types, for example origin-destination estimation based on location area updates.