From classical to quantum machine learning: survey on routing optimization in 6G software defined networking

Abstract

The sixth generation (6G) of mobile networks will adopt on-demand self-reconfiguration to fulfill simultaneously stringent key performance indicators and overall optimization of usage of network resources. Such dynamic and flexible network management is made possible by Software Defined Networking (SDN) with a global view of the network, centralized control, and adaptable forwarding rules. Because of the complexity of 6G networks, Artificial Intelligence and its integration with SDN and Quantum Computing are considered prospective solutions to hard problems such as optimized routing in highly dynamic and complex networks. The main contribution of this survey is to present an in-depth study and analysis of recent research on the application of Reinforcement Learning (RL), Deep Reinforcement Learning (DRL), and Quantum Machine Learning (QML) techniques to address SDN routing challenges in 6G networks. Furthermore, the paper identifies and discusses open research questions in this domain. In summary, we conclude that there is a significant shift toward employing RL/DRL-based routing strategies in SDN networks, particularly over the past 3 years. Moreover, there is a huge interest in integrating QML techniques to tackle the complexity of routing in 6G networks. However, considerable work remains to be done in both approaches in order to accomplish thorough comparisons and synergies among various approaches and conduct meaningful evaluations using open datasets and different topologies.