团队文章“An Edge Computing Based Public Vehicle System for Smart Transportation" 被 IEEE Transactions on Vehicular Technology.”接收。

Transportation-based cyber-physical system (TCPS), also known as intelligent transportation system (ITS), is one of new and important applications and trends in transportation. With integrating communication protocol, real-time system, interconnection networks and transportation management and control techniques, TCPS can make transportation systems reliable, efficient, and secure. To increase traffic efficiency, a number of public vehicle systems and ridesharing schemes have been developed to prompt travelers to share rides during their trips, thereby reducing the number of vehicles on roads. Nonetheless, existing schemes do not consider the traveling satisfaction in the view of travelers when matching ride requests to vehicles. In addition, in existing schemes, the ride requests are mostly processed in data center (or cloud), which would bring large computation overhead in data center, resulting in long latency of ride demand responses. To this end, in this paper we propose an Edge Computing-Based Public Vehicle System (ECPV) to improve traffic efficiency and vehicle occupancy rate by scheduling public vehicles and sharing rides among travelers with considering the satisfaction of travelers. Particularly, ECPV considers public vehicle scheduling (i.e., ridesharing) as an optimization problem with the objective of maximizing the satisfaction of travelers, in which the satisfaction jointly considers traveling distance, traveling time and traveling charge to achieve a comprehensive reflection of travelers' preferences. Due to optimization problem is a NP-hard problem, an edge computing based ride requests transmission mechanism and a tree-based heuristic matching mechanism are proposed to transmit ride requests launched by travelers and select appropriate vehicles for ride requests, in which most of ride requests can be responded and matched to vehicles in edge devices, resulting in reduction of computation overhead in data center. In addition, a graph partition based depot placement mechanism is proposed as well to determine the appropriate locations for vehicle parking so that vehicle distribution can be balanced over the whole transportation system. Through an extensive performance evaluations, our experimental results show that our proposed ECPV system can effectively match ride requests to public vehicles and reduce the travel time and travel distance for all trips, reduce waiting time and travel charge of travelers, improve vehicle occupancy rate and traffic efficiency, as well as achieve great traveler satisfactions.