Abstract
In-advance availability of parking information plays an important role in parker/ traveller decision-making for parking, curbing congestion, and managing parking lots efficiently. Specifically on-street parking poses many challenges compared to the off-street ones. Many users such as, store owners, municipal authorities, and police demand slots for on-street parking Free of Charges (FoC) for a short duration. In last few years, parking authorities collected data to attract the attention of researchers to present data-centric solutions for various problems such as, minimization of parking prices, maximization of revenue, and balacing the congestion at parking lots associated with smart parking systems. Motivated from the aforementioned problem, this paper proposes a scheme based on machine learning and game theory for dynamic pricing and allocation of parking slots in on-street parking scenarios. The dynamic pricing and allocation problem is modeled as Stackelberg game and is solved by finding its Nash equilibrium. Two types of Parking Users (PUs), i.e., Paid Parking Users (PPUs) and Restricted Parking Users (RPUs) are considered in this work. RPUs avail parking slots FoC once a day. PPUs compete to minimize the prices, and RPUs compete to maximize the FoC granted duration. The Parking Controllers (PCs) compete to maximize revenue generated from PPUs and to minimize total FoC parking duration granted to the RPUs. The random forest model is used to predict occupancy, which in turn is used to generate parking prices. Seattle city parking and its prices data sets are used to predict occupancy and to generate prices, respectively. In order to test the performance of communication system, the proposed DyPARK Pricing and Allocation Scheme (PAS) is compared with its four variants and is found worth. The proposed scheme is also compared with other state-of-the-art schemes using various performance evaluation metrics. Simulated results prove the superiority of the proposed scheme in comparison to the other state-of-the-art schemes.