Prediction-based parking allocation framework in urban environments

Abstract

ABSTRACT Finding a parking space is usually challenging in urban areas. The literature shows that 30% of traffic congestion is caused by searching for parking spaces, which results in unnecessary energy consumption and environmental pollution. With the development of sensor technologies, smart parking guidance systems provide users with a variety of real-time parking space information. However, users cannot know whether the target parking space remains available upon arrival. Moreover, parking resources may be under competition when multiple users target the same open parking space. In this research, we develop a new framework named prediction-based parking allocation (PPA) that provides smart parking services to users. In PPA, we first construct a prediction model of parking occupancy and predict the subsequent parking availabilities. Then, we design a matching-based allocation strategy to assign users to selected parking spaces. To the best of our knowledge, this is the first study that combines occupancy prediction and space allocation simultaneously to address smart parking issues. Finally, we collect a real dataset from the SFPark on-street parking system for performance evaluation. According to experimental results, PPA can effectively increase the parking success rate and reduce costs, fuel consumption, and carbon emissions.