On the Serviceability of Mobile Vehicular Cloudlets in a Large-Scale Urban Environment

Recently, cloud computing technology has been utilized to make vehicles on roads smarter and to offer better driving experience. Consequently, the concept of mobile vehicular cloudlet (MVC) was born, where nearby smart vehicles were connected to provide cloud computing services locally. Existing researches focus on MVC system models and architectures, and no work to date addresses the critical question of what is the potential, i.e., level of local cloud computing service, achievable by MVCs in real-world large-scale urban environments. This issue is fundamental to the practical implementation of MVC technology. Answering this question is also challenging because MVCs operate in highly complicated and dynamic environments. In this paper, we directly address this challenging issue and we introduce the concept of serviceability to measure the ability of an MVC to provide cloud computing service. In particular, we evaluate this measure in practical environments through a real-world vehicular mobility trace of Beijing. Using the time-varying graph model for mobile cloud computing under different scenarios, we find that the serviceability has a relationship with the delay tolerance of the undertaken computational task, which can be described by two characteristic parameters. The evolution of serviceability through a day and the influence of network congestion are also analyzed. We also portray the spatial distribution of the serviceability and analyze the influence of connectivity and mobility in both MVC and vehicle levels. Our observations are valuable to assist designing vehicular cloud computing systems and applications, as well as to help make offloading decisions.