Predictive Task Migration Modeling in Software Defined Vehicular Networks

Fog computing has promoted the rapid development of the Internet of Things (IoT), where task offloading is becoming one of the most achievable approaches to solve the limitation of on-board computing resources. However, in practical applications of intelligent transportation systems (ITS), the task offloading scheme lacks adaptability to rapidly changing dynamic network topologies. In this paper, leveraging the idea of logical centralized control in software defined networks (SDN), we investigate the optimization problem of fog node resource allocation in vehicular task offloading scenarios aiming to minimize the latency and energy usage. Specifically, we propose a resource migration method for real-time vehicle tasks to avoid redundant downloads of data through configuring fog node resources. Meanwhile, we build a predictive dynamic task migration model to evaluate our system by calculating the delay and energy consumption. Finally, the effectiveness of the model proposed in this paper is validated through a case study with processing of real-time road perception applications in a software defined vehicular network.