Toward Just-in-Time Data Communications over Shared Networks and Computational Resources on Massive Client Environment

Abstract

Intelligent driving has been benefiting from advances in automotive big data analysis and on-demand data communications, where efficient vehicle-to-cloud communication is a key technology able to collect a huge amount of data from vehicles. However, as the capacities of the network and computational resources of the cloud are not unlimited, simultaneous data transfer based on a best-effort manner results in a resource starvation problem. Therefore, an application is needed to prioritize data communication to alleviate the problem and complete the data transfer just in time. This paper highlights the problem of resource starvation led by best-effort data transfers. It proposes protocols for data communication and control that could enable just-in-time data communication, notably for data collection. We demonstrate through experiments that the proposed mechanism enables efficient and effective data collection over the shared network resources.