A Multi-objective Optimization Approach for SDVN Controllers Placement Problem

The Software Defined Networking (SDN) paradigm consist of decoupling the control from the data plane. Recently, the adoption of the SDN paradigm as the basic architecture for Vehicular networks (SDVN) coupled with the 5G promises to accelerate the Intelligent Transport Services and smart cities deployment. However, it raises many challenges generated mainly by the dynamic nature of the vehicular network and the centralized aspect of the control plane. The distributed control plane has been identified as suitable architecture for such environment. Hence, this study focuses on the SDVN Controller Placement Problem (CPP). Previously, several researches addressed this problem in the context of wired networks by considering primary metrics such as control path latency and controller capacity. In this paper, we propose to adopt a multi-objective optimization approach to elect the nodes designated as controllers. The election is done by considering different conflicting metrics: number of controllers, latency, load balancing metric and a key metric in distributed system, i.e: clock offset between the controllers and the vehicular network nodes for controllers synchronization. The multi-objective genetic algorithm is used to solve this multi-objective optimization problem and create a compromise controllers placement solution. Two topology models have been considered to evaluate the performances. The analysis of the simulation results shows the feasibility of our algorithm. The simulation gives promising results in both scenarios.