Abstract: In a ridesharing scenario, multiple passengers with similar itineraries and time schedules take one vehicle to reduce travel cost, increase vehicle occupancy and ease traffic congestion. However, the existing studies ignore the effect of inequitable charging standard and malicious bidding behavior on passengers ridesharing Quality of Experience (QoE). Considering the constraints of cost, vehicle capacity and detour distance , this paper proposes the problem of maximizing the fairness of matching results, and models the process of ridesharing pricing and matching as a two-stage Stackelberg game. Aiming at the above problems, a request division algorithm based on K-means++ is proposed to narrow the match- ing range and improve the matching efficiency. On the premise of satisfying all participant constraints, an iterative algorithm DPMA based on two-stage Stackelberg game is designed and its convergence is proved theoretically. Simulation experiments are carried out on the New York taxi dataset, and the convergence of the algorithm DPMA is verified under different parameter settings. Compared with two existing algorithms, DPMA improves the fairness index by 34.03% and 24.42%, respectively, while ensuring the drivers benefits. The experimental results show that the designed mechanism can effectively avoid malicious bidding among drivers and improve the fairness of ridesharing matching.

Key words: ridesharing, Stackelberg game, fairness