Deployment and Dimensioning of Fog Computing-Based Internet of Vehicle Infrastructure for Autonomous Driving

Internet of Vehicle (IoV) is an important paradigm to realize the intelligent transportation system. However, with the increasing number of vehicular applications, how to satisfy the ubiquitous requirements of communication and computation is challenging. Fog computing provides the real-time transportation services to local users timely through close-proximity data processing, rather than routing data to a remote central data center in the cloud. More importantly, the fog computing will facilitate the autonomous driving (AD) revolutionarily. This paper investigates the problem of optimal deployment and dimensioning (ODD) of fog computing-based IoV infrastructure for AD. For the ODD problem, we present two diverse architecture modes, i.e., the coupling mode (CRF) and the decoupling mode (DRF), and formulate the ODD problem into two integer linear programming formulations with the objective of minimizing the deployment cost. A heuristic algorithm is also proposed to achieve the suboptimal deployment solution for large-scale fog computing-based IoV. Numerical results show that the DRF is more cost-effective and flexible than the CRF for deployment in practice.