A throughput-aware joint vehicle route and access network selection approach based on SMDP

In intelligent transportation system (ITS), the interworking of vehicular networks (VN) and cellular networks (CN) is proposed to provide high-data-rate services to vehicles. As the network access quality for CN and VN is location related, mobile data offloading (MDO), which dynamically selects access networks for vehicles, should be considered with vehicle route planning to further improve the wireless data throughput of individual vehicles and to enhance the performance of the entire ITS. In this paper, we investigate joint MDO and route selection for an individual vehicle in a metropolitan scenario. We aim to improve the throughput of the target vehicle while guaranteeing its transportation efficiency requirements in terms of traveling time and distance. To achieve this objective, we first formulate the joint route and access network selection problem as a semi-Markov decision process (SMDP). Then we propose an optimal algorithm to calculate its optimal policy. To further reduce the computation complexity, we derive a suboptimal algorithm which reduces the action space. Simulation results demonstrate that the proposed optimal algorithm significantly outperforms the existing work in total throughput and the late arrival ratio. Moreover, the heuristic algorithm is able to substantially reduce the computation time with only slight performance degradation.