ABSTRACT
Nowadays, the flourishing vehicular services generally have severe Quality-of-Service (QoS) requirements in Internet of Vehicles (IoV), making cloud-based processing architectures infeasible. Therefore, Edge Computing (EC) emerged as a critical component by its physical proximity to information-generation vehicles. Although many measures are adopted for resource optimization in IoV, they hardly enable intelligent decisions while satisfying real-time service requirements. With the revival of Artificial Intelligence (AI), we can push the AI frontiers to the network edge in IoV, which gives rise to an emerging interdisciplinary, Edge Intelligence (EI). EI can sink the cloud’s processing capabilities to the edge side in IoV, thus assisting in resources optimization adaptively. However, EI empowered resource optimization in IoV is still in its infancy. This chapter aims to provide a study with relative concepts and prospects of this young field from a broader perspective. We first describe the background, including the concept of IoV and EI. Then, we introduce the EI-supported IoV architecture, based on which we make a holistic overview of its related key enablers and specific optimization themes. Subsequently, the processes of model training and inference in IoV are elaborated, together with its particular embodiment of resource optimization. Finally, we strive to articulate open challenges and promising research directions, which may facilitate the transformation of EI empowered resource optimization in IoV from theory to practice.
