Efficient and privacy-preserving multi-agent systems for smart city carpooling with k-regret queries and differential privacy

Abstract

Multi-Agent Systems are characterized by the presence of multiple independent agents and find diverse applications. In the context of smart cities, MAS is employed in traffic management to enhance operational efficiency, optimize resource utilization, and improve the quality of life for residents. This research paper focuses on the design of a multi-agent intelligent scheduling system, where passengers, vehicles, and carpooling platforms serve as intelligent agents. The primary objective of passengers is to identify suitable shared vehicles based on criteria such as waiting time, budget constraints, and willingness to carpool. Vehicles, on the other hand, organize their schedules based on passenger demands and designated routes. The carpooling platform takes into account resource allocation priority and optimization problems to ensure the efficient operation of the system. To address the issue of vehicle ordering, k-regret queries are utilized, while passenger preferences provide insight into determining loss factors. To safeguard privacy, differential privacy techniques and a random response mechanism are employed when dealing with multiple passenger queries. Furthermore, a direction-preserving insertion verification method is implemented to mitigate computational complexity. The effectiveness and efficiency of the proposed approach are validated through experimentation.