Improving Energy Consumption in Software Defined Network by Predicting Optimized Flow Routing Using Deep Learning Method

Currently Software Defined Network (SDN) plays a major role in Data Centers (DCs) and widely used.These networks programmable has influenced the better features which admit innovation in deploying the enormous recent applications in faster and secured manner.This advancement exhibit with cost of high processing power as well as energy consumption.Several researchers have undertaken these problems by existing routing techniques in dynamic adjusting of forwarding plane network for saving energy.In accomplishing fine grained network performance through optimizations for flow routing using different routing packets traffic flow with distinct network paths.Implementation of centralized network optimizations can be controlled through SDN centralized controller by optimized flow routing.The flow routing implementation is adjustable by altering traffic loads for complex models.However, the study goal is in pursuing a model associated with Reinforcement Learning (RL) for rerouting the flow of SDN.Therefore, this paper introduced RL based Convolution Neural network (CNN) with hyperparameter modelling for better performance of energy efficiency of the network by improving the already proposed route selection.The proposed RL-CNN with hyperparameter model has provided a suitable Quality of Service (QoS) over hybrid IP or SDNs which assist in coordinating both IP and SDN paradigm.Traffic balancing using minimum network power consumption and link utilisation over hybrid IP or SDNs with assist of proposed RL-CNN with hyperparameter.The proposed RL-CNN is assessed over topologies of various sizes utilizing various techniques to determine which nodes need to be converted from IP to SDN.