Efficient Load Balancing in Software Defined Networks Using Unsupervised Meta-Reinforcement Learning

New technologies, such as cloud computing and big data, make traditional network management more difficult, while the network's user base continues to expand. Consequently, changes to the traditional network architecture are required. To address this issue and enhance network management compliance, a new idea called software-defined network (SDN) has been proposed. Due to limited network resources and to satisfy the criteria of quality of service, it is necessary to address numerous difficulties, one of which is load balancing (LB), which distributes data traffic among several resources to maximize their efficiency and reliability. The local information of a network is used to design LB in a typical network. Consequently, it is not very accurate. Still, software-defined network (SDN) controllers can see the network from anywhere in the world and make better LB. We provide a set of unsupervised meta- RL algorithms to build smart optimization architecture for SDN LB.