SDN-Based Internet of Autonomous Vehicles: An Energy-Efficient Approach for Controller Placement

摘要

The rapid advancement of the Internet of Things is expected to play a critical role in future intelligent transportation systems. This technology utilizes advanced information and communication technologies to enhance the operational capabilities of the vehicles and is often referred to as IoAV. More importantly, the increasing usage of sensors and other technologies generates vast amounts of data and information to be exchanged between AVs. Since the wireless connectivity between AVs is constantly expanding, the transmission of data is expected to pose several challenges to the conventional wireless networks, including resource utilization, network optimization, quality of service, and so on. To overcome these challenges, software-defined networking (SDN) has emerged as a powerful technology. This article presents a composite architecture named SD- IoAV for the integration of SDN with IoAV. This integration is not straightforward because SD-IoAV is geographically dispersed by nature. An efficient technique to manage the underlying communications is to deploy multiple SDN controllers across the widely dispersed SDN domains. This is referred to as the controller placement problem (CPP). Thus, the primary focus of this work is to explore CPP in the context of SD-IoAV as a special case of energy minimization and load balancing under latency restrictions. In this context, for large networks, the number of variables increases exponentially, which in turn escalates the complexity of the problem manifold. Hence, to deduce a near optimal solution for large networks, a heuristic approach based on the incremental expansion of candidate space is proposed. The simulation results have been carried out in MATLAB, and the obtained results show that the proposed scheme attains higher energy savings and better load capacity management compared to an existing technique, that is, improved performance by 18.73 and 9.42 percent, respectively.