Estimation of time-dependent travel time in an urban network is a challenging task due to the interrupted nature of vehicular traffic flows. A novel concept of intersection-to-intersection (I2I) real-time travel time estimation (TTE) and route suggestion model based on vehicular ad-hoc network (VANET) technology is proposed for enabling the smart transportations in smart city. The system components, communication protocol and collaborative intelligence concept are designed to facilitate real-time vehicular applications. Vehicles equipped with an on-board unit (OBU) send TTE requests to the road side unit (RSU) and share their real-time information, including traveled path and average speed, and the RSU responds with the suggested shortest route as well as the estimated travel time. In order to efficiently share real-time traffic information among RSUs, a propagation-based RSU-to-RSU (R2R) data exchange algorithm and a traffic information super-matrix data structure are designed. These reduce the complexity from O(N2) for a traditional broadcast approach to O(N). Real data collected from a GPS-based taxi dispatching system is applied to evaluate the accuracy of the proposed TTE model and the performance of the suggested route. The experimental results show that the average mean absolute error percentage (MAPE) of the proposed TTE model is 13.6% compared to real taxi journeys, which indicates that the performances of the suggested routes are good. The TTE of the suggested paths has the possibility of being 82.2% better than the paths traveled by taxi, and the travel time is thus reduced by 15.9% on average over a year.
All Science Journal Classification (ASJC) codes