Heterogeneous traffic information provision on road networks with competitive or cooperative information providers

Abstract

This paper studies the long-term effects of heterogeneous information provision on road networks with multiple information providers (IPs). To capture the information heterogeneity resulting from various IPs, we consider that users subscribing to different IPs receive non-identical information sets about the available routes and make route choices within their information sets. We investigate both IPs’ strategic interactions and users’ route choices in a bi-level “leader–follower” game. At the upper level, IPs act as leaders who compete or cooperate to determine the optimal information sets provided for their users to maximize their objectives. Three models are developed to investigate IPs’ strategic interactions, i.e., Nash game, Stackelberg game, and cooperation, in which IPs are non-cooperative, leader and followers, and cooperative, respectively. At the lower level, users act as followers to choose their user-optimal routes based on their received information sets. Wardrop Equilibrium with Multiple Information Classes (WEMIC) is defined to capture users’ route choices. The WEMIC problem is formulated as an equivalent variational inequality (VI) problem. The existence and uniqueness of the VI solution are proved. When IPs play the Nash game, the problem is formulated as an Equilibrium Problem with Equilibrium Constraints (EPEC) with discrete variables and non-continuous payoff functions. To solve the EPEC efficiently, we propose a Binary Coded Nash Dominance based Evolutionary (BCNDE) algorithm embedded with a heuristic method for action space reduction. Finally, numerical experiments with two IPs (a big IP and a small IP) are conducted to analyze the effects of unit information cost in the Nash game and the impact of the leader’s strategy in the Stackelberg game, respectively. We find that the small IP may benefit from higher information costs in the Nash game. Moreover, the effects of IPs’ competition and cooperation are compared from both the perspectives of IPs and the system. The results indicate that cooperation can reduce congestion and make both IPs better off. On the Nguyen–Dupuis network, we demonstrate the effectiveness of the BCNDE algorithm and analyze the effects of market share distribution in the Nash game. We also extend this study by considering elastic demand. The extended models, theoretical proofs, and corresponding numerical experiments are provided.