Machine intelligence approach: To solve load balancing problem with high quality of service performance for multi-controller based Software Defined Network

摘要

In this paper, we have proposed a DRL based method to obtain the route based on an optimized load of SDN which is based on self-learning of human intelligence. In this proposal, the Bio-Inspired RBM is used for BioInspired Deep Belief Architecture (BDBA) for implementing deep learning to obtain the optimized route. This Bio-Inspired RBM has two parts one is simple RBM and another part is inspired by self-learning of human intelligence based on emotion learning of the limbic system of the brain. Every Bio-Inspired RBM is fined tune using the reward function R which captures the environmental dynamics in the form of network policies.Software Defined Network (SDN) concept resolves several problems of network infrastructure to decouple the responsibilities of the control plane and data plane. The single controller improves control on the network but decreases the reliability of the system in the case of failure of the controller. The distributed controller improves reliability and also reduces system failure. The route optimization is a big challenge in distributed SDN. Some route optimization techniques have been proposed which requires some prior knowledge. Deep Reinforcement Learning (DRL) is one of the techniques for route optimization which does not require any prior knowledge and runs in real-time. This technique learns from environment dynamics and optimizes anonymously. The high demand for internet usages and business activities using distributed SDN also facing complex problems that can be resolved using the self-learning methodology of human intelligence. Self-learning of human intelligence plays a dominant role in making complex decisions for any sudden problem.