A Performance Evaluation of Vehicular Swarm Intelligence for Seamless Route Guidance using Opportunistic Networking

We considered a vehicular route guidance system in normal times that can be seamlessly transferable to a evacuation guidance system in times of disaster. In recent years, based on the high penetration rate and availability of smartphones, several swarm intelligence systems have been proposed where mobile nodes autonomously collect information on impassable and congested areas by disaster using their smartphone applications to guide evacuation by sharing the information with other nodes in proximity. It was not fully understood whether this guidance method in times of disaster is also effective as a route guidance method in normal times. For example, the impact of the method on travel time and distance have not been theoretically evaluated in normal times, where there are various destinations for each traffic vehicle. In addition, it was not known what proportion of the vehicles with proximity information sharing will be effective for smooth movement in normal times. In this study, we performed simulations of vehicles and their opportunistic networking in grid-like road networks assuming an urban area to investigate the effects of the number of traffic vehicles, the proportion of traffic vehicles sharing proximity information, and the number of road closures on distributions of travel time and distance. By using Wilcoxon signed-rank test, we found scenarios in which there was a significant difference in the distributions compared to the corresponding scenarios without proximity information sharing. As a result, we confirmed that if the proportion of vehicles sharing proximity information is more than 50%, significant differences in the distributions occur when the size of the road network and the number of vehicles are sufficiently large, where a large-scale traffic congestion is more likely to occur. We also confirmed more significant differences as increasing the number of road closures.