MobiPlace: Mobility-Aware Controller Placement in Software-Defined Vehicular Networks

Abstract

In this paper, we propose a mobility-aware scheme, named MobiPlace, to address the controller placement problem (CPP) at the Road Side Units (RSUs) in Software-Defined Vehicular Networks (SDVNs). MobiPlace places local controllers at the selected RSUs to reduce the operational delay experienced in traditional SDVN architecture, where controllers are placed at the cloud. Additionally, we consider the effect of dynamic road traffic and propose dynamic adjustment of the placement of the controller with minimal changes. In contrast to the existing literature, we infuse traffic monitoring and traffic prediction strategies to ensure accurate delivery of control messages and data packets. We formulate an Integer Linear Program (ILP) and propose a solution approach consisting of three modules — mobility management , controller placement , and controller selection . The mobility management module uses Markov predictor to predict vehicle movement and reduce controller synchronization overhead when a vehicle moves to a different controller's coverage area. The controller placement module applies a simulated annealing-based algorithm to select potential RSUs that serve as local controllers. The controller selection module determines the preferable controller location for processing the service requests. Simulation results depict that MobiPlace reduces the average flow setup delay by $13.85 \%$ compared to the existing state-of-the-art.