ParkingRank-D: A Spatial-temporal Ranking Model of Urban Parking Lots in City-wide Parking Guidance System

City-wide parking guidance systems (CPGS) is a specialized cyber-physical search engine for vehicles to look for proper available parking lots nearby, aiming to relieve the parking pains. If from the perspective of search engine, the parking lots will constitute a city-wide network, parking lots as nodes and roads linking them as edges. Inspired by PageRank, the classic ranking model in WEB, we propose a new ranking model named ParkingRank-D, to evaluate the importance of parking lots all over the city, and help recommend to vehicles more accurately. ParkingRank-D is a spatial-temporal ranking model, including three steps. Firstly, building the quantified service capability model of parking lots as the initial ranking value. But unlike PageRank, the transition probabilities among parking lots are not inherently equal, but depend on the distance and the number of parking spaces available when vehicles arriving. So, next we build a transition matrix according to the geospatial relations and the spaces available of the parking lots. The final step is iteration, till the all ranking values in convergence. The experiment shows, as the important underlying technology for CPGS, ParkingRank-D seems more complicated, but it can rank the target parking lots more precisely from spatial-temporal perspective. The ranking results are easily understood and accord with the people’s cognition of parking possibility very well.