Cooperative Task-Oriented Group Formation for Vehicular Networks

As vehicles are embedded with an increasing number of sensors and more powerful processors, computation-intensive on-board applications are being deployed. Emerging cooperative processing capabilities among vehicles will increase computing capability even further. In this paper, we present a novel framework for task-oriented group formation, where groups of vehicles are tailored for a specific cooperative computation task to be performed. We use the framework to develop a vehicular group formation algorithm that improves the quality of the computation result while achieving a specified probability of successful task completion. A prototype of the group formation algorithm for a generic distributed learning application example is implemented and extensively evaluated. Results show that our approach is able to significantly increase the percentage of successfully completed tasks compared to two baseline approaches.