Development and implementation of a game theory based ride-sharing technique

• Main Authors: Borja, Gabriel Antonio M., Ching, Gerard Ryan C., Espiritu, Francis Miguel M., Go, Kerwin D.
• Format: text
• Language: English
• Published: Animo Repository 2022
• Subjects: Game theory; Ridesharing; Global Positioning System; Computer Engineering
• Online Access: https://animorepository.dlsu.edu.ph/etdb_ece/7; https://animorepository.dlsu.edu.ph/cgi/viewcontent.cgi?article=1005&context=etdb_ece
• Institution: De La Salle University

Mobile internet technologies have sparked a multitude of opportunities for people to interact with one another and share resources. These have paved way to the rise of a phenomenon known as the “sharing economy” defined as sharing the resources through the internet. An application of sharing economy is ride-sharing where drivers offer their vehicles as a mode of public transportation to multiple passengers. In this work, we propose a Game Theory-based solution to address the stable matching among riders while minimizing their cost. Two stable matching techniques are proposed in this study, namely: First-Come, First-Served (FCFS) and Best Time Sharing (BT). FCFS discovers pairs based on earliest time of pair occurrences, while BT prioritizes selecting pairs with high proportion of shared distance between passengers to the overall distance of their trips. We evaluate our methods through extensive simulation from empirical taxi traces from Jakarta, Singapore, and New York. Results in terms of of post- stable matching cost savings, travel distance, successful matches, running trips, and spatio-temporal distribution have been evaluated to gauge the performance with respect to the no ridesharing condition. BT outperformed FCFS in terms of generating more pairs with compatible routes. Additionally, in the NY dataset with high amount of trip density, BT has efficiently reduced the number of trips present at a given time. On the other hand, FCFS has been more effective in pairing trips for the JK and SG datasets because of lower density due to limited amount of trajectories. The Game Theory (GT) pricing model 5 proved to generally be the most beneficial to the ride share’s cost savings, specifically leaning toward the passenger benefits. Analysis has shown that the stable matching algorithm reduced the overall number of trips while still adhering to the temporal frequency of trips within the dataset. Moreover, our developed Best Time Pairing and Game Theory Pricing methods served the most efficient based on passenger cost savings. Applying these stable matching algorithms will definitely benefit more users and will encourage more ridesharing instances.