Designing a dynamic carpooling system integrated with the VANET-based route-planning algorithm

It is known that most traditional carpooling approaches fail to satisfy all carpooling constraints and fail to provide the best carpooling sequence to users within a reasonable duration of computing time. In recent years, a vehicle has become able to exchange its real-time traffic information with nearby vehicles via the vehicular ad hoc network (VANET) to re-calculate its path to the destination so as to save time and fuel. Here, we propose a dynamic Carpooling System with real-time Vehicular Information (CSVI), including expounding its system architecture, message flows, and carpool-matching algorithms. CSVI not only integrates with the VANET-based A* (VBA*), which is a VANET-based route-planning algorithm that we have previously proposed, but also computes carpooling sequences to minimize fuel costs by speeding up with three algorithms to satisfy four carpooling constraints. Furthermore, using historic vehicle density statistics of Taipei City, Taiwan, we perform simulations on three performance metrics versus three carpooling parameters to evaluate carpooling efficiencies of six route-planning and carpool-matching mixtures. According to these simulation results, CSVI is the best system integrating these metrics and parameters. CSVI significantly outperforms traditional carpooling approaches and consumes much less computing time than the optimal carpooling approach.