Improving sensing coverage of probe vehicles with probabilistic routing

Modern cars are pervasively equipped with multiple sensors meant to improve in-vehicle quality of life, efficiency and safety. The aggregation on a remote back-end of the information collected from these sensors may give rise to one of the biggest and most pervasive sensor networks around the world, making possible to extract new knowledge, or contextual awareness, in a detail never experienced before. Anyhow, an open issue with probe vehicles is the achievable spatio-temporal sensing coverage, since vehicles are not uniformly distributed over the road network, because drivers mostly select a shortest time path to destination. In this paper we present an evolution of the standard A* algorithm, where the route is chosen in a probabilistic way, with the goal to maximize the spatio-temporal coverage of probe vehicles. The proposed algorithm has been empirically evaluated by means of a public dataset of more than 320.000 real taxi trajectories, showing promising performances in terms of achievable sensing coverage.