DESIGN AND DEVELOPMENT OF REAL TIME AUTOMATED ROUTING ALGORITHM FOR COMPUTER NETWORKS

DESIGN AND DEVELOPMENT OF REAL TIME AUTOMATED ROUTING ALGORITHM FOR COMPUTER NETWORKS In computer networks, efficient usage of network resources is by logically adapting centralized control where the control plane is decoupled from data plane. Conventional routing strategies rely on manual configuration which results in degraded performance of the network especially Software Defined Network (SDN) suffer from inefficient allocation of bandwidth due to link congestion in data flow. In this invention, we propose novel network architecture driven intelligently for automatic routing. The proposed automated routing algorithm adapts deep reinforcement learning for simulation of human learning knowledge methods for controlling the closed loop mechanism incorporated with technologies for network monitoring for realizing the network environment interaction. This work interacts with network environment for better control decision by adaptive learning. Resource availability and services related are adjusted intelligently by the network based on environmental conditions. Deep reinforcement learning based automatic routing algorithm finds the nearest optimal path in real time network environment hence efficient in routing of computer networks. State A cio Network vDynamic and Stegy peiamance update Control C) SDN Controller planeJ Information _ _ _ Flo leb collection- - - - - n albo Forwarding plane Figure 1. Proposed Architecture of Autonomous Routing Algorithm for Computer Networks Samples Replay Random Sample Environment v9(s,alse) s~-SVrp~la)U Actor - tPa p )ricnty Critic Gradie g FOne Figure 2. Deep Deterministic Policy Gradient based Autonomous Routing Algorithm for Computer Networks