Online Data Traffic Steering in Software-Defined Autonomous Vehicle Networks

Abstract

In the past decade, autonomous driving technologies have experienced a significant growth. In order to meet the increasing data transmission demands from autonomous vehicles (AVs), a novel network paradigm connecting AVs with the Internet is needed. In this paper, we first present the Software-Defined Autonomous Vehicle Networks (SD-AVN) framework to bridge the gap by introducing Software Defined Networking (SDN) and fog computing technologies. With SDN, we focus on a centralized routing problem in SD-AVN, and our intent is to minimize the overall transmission cost by reducing the usage of 5G base stations (BSs). Motivated by this, we formulate the global routing problem as a mixed integer programming (MIP) problem and develop an online log-competitive approximation algorithm to solve it. After that, we also explain that the computation-intensive routing tasks can be distributed to different fog controllers to reduce the scheduling time and end-to-end delay. Experimental results validate the effectiveness of the proposed algorithm in comparison with other two routing heuristics.