A utility-based resource allocation scheme in cloud-assisted vehicular network architecture

In the era of the Internet-of-Vehicles (IoV), all components in an Intelligent Transportation System (ITS) can be connected to improve the traffic safety and transportation efficiency. In order to maximize the utilization of the resources, e.g., computation, communication and storage resources, the cloud computing technique could be integrated into vehicular networks. Meanwhile, a cloud-assisted vehicular network could be proposed for effective resource management. In this paper, the resource allocation problem in the cloud-assisted vehicular network architecture is investigated. Each cloud in the architecture has its own specific features, e.g., the remote cloud has sufficient resources but experience high end-to-end delay while the local cloud and vehicular cloud have limited resources but a lower transmission delay is attained. The optimal problem to maximize the system expected average reward is formulated as a Semi-Markov Decision Process (SMDP). Consquently, the corresponding optimal scheme is obtained by solving the SMDP problem via an iteration algorithm. The proposed scheme can provide guidelines that will be helpful to decide to which cloud a request should be admitted and how many resources are needed to be allocated. Numerical results indicate that the proposed scheme outperforms other resource allocation schemes and improves system rewards as well as the obtained experience by the vehicular users.