Traffic-aware Dynamic Controller Placement using AI techniques in SDN-based aeronautical networks

Abstract

In software-defined networks (SDNs), multiple distributed controllers have been used to improve flexibility, scalability and reliability, where the controller placement remained fixed over time. However, the mobility of aircrafts complicates Aeronautical Telecommunication Network (ATN) operations; aircraft mobility results in time-varying and nonuniform geographical aircraft distribution. Given the dynamics of traffic patterns in the ATN network, a fixed solution of the Controller Placement Problem (CPP) will not be accurate. In this paper, we show that dynamic controller placement and dynamic switch-to-controller assignment can improve system elasticity and efficiency in managing traffic load variations. Toward this end, the traffic-aware controller placement (TACP) problem has been formulated as an Integer Linear Program (ILP) to achieve optimal load balancing between controllers. An optimal enumeration based algorithm called Dynamic Placement Fastest Assignment (DPFA) is proposed. Further, an Artificial Intelligence (AI) algorithm called Genetic Algorithm based Dynamic Placement Dynamic Assignment (GA-DPDA) is developed to achieve the near-optimal load balancing performance. Simulations demonstrate that DPFA and GA-DPDA significantly reduce the load imbalance among SDN controllers.