Development of a Real-Time Model of the Occupancy of Short-Term Parking Zones

The lack of reliable information about the occupancy of on-street parking places in the inner districts of large cities causes an unnecessarily high amount of parking search traffic. Previous attempts to solve the problem rely on accurate information systems that navigate drivers to the next available parking space. They have failed due to high costs for roadside sensors and are also controversial, because they indirectly support car use. A service that informs road users early about the situation at the destination would support a shift towards public transport modes and reduce parking search traffic particularly at times of unexpectedly high parking pressure. The real-time occupancy model presented in this article does not aim to indicate the occupancy of single parking lots, but to provide a sufficiently accurate description and reliable prediction of the occupancy at the destination. Three existing real-time indicators were tested for their predictive power: (i) location data of mobile-phone parking; (ii) counts of car parks in the vicinity; and (iii) traffic flow volumes. Experimental results indicated that models using mobile-phone parking data have the potential to improve predictions of the occupancy rate by detecting exceptional deviations from the average day curve.