Intelligent traffic control for QoS optimization in hybrid SDNs

Abstract

Software Defined Networking (SDN) provides a flexible way to control traffic in networks and it is seen a rapid increase among network operators in adoption of SDN. However, due to some policy and economic issues, the coexistence of SDN-enabled devices and legacy devices will continue for a long time. This hybrid scenery comes with many challenges that do not exist in pure SDN. How to find an efficient and suitable routing policy in a hybrid SDN is essential for promoting the development of SDN. In this paper, we propose a near-optimal traffic control method for QoS optimization in a hybrid SDN. First, an SDN migration sequence is explored to maximize controllable traffic to improve the effects of optimization. Then, a Deep Reinforcement Learning (DRL) algorithm is used to address the multi-splittable routing problem in the hybrid SDN. The flow split ratio strategy is implemented by setting the OpenFlow group bucket constraints. Finally, we evaluate the proposed method with open-source traffic datasets. The simulation results show that the method of this paper can achieve a significant improvement in optimizing network QoS performance such as delay, jitter, and link utilization.