Reliability Awareness Multiple Path Installation in Software Defined Networking using Machine Learning Algorithm

Abstract

Link failure is still a severe problem in today's networking system. Transmission delays and data packet loss cause link failure in the network. Rapid connection recovery after a link breakdown is an important topic in networking. The failure of the networking link must be recovered whenever possible because it could cause blockage of network traffic and obstruct normal network operation. To overcome this difficulty, backup or secondary channels can be chosen adaptively and proactively in SDN based on data traffic dynamics in the network. When a network connection fails, packets must find a different way to their destination. The goal of this research is to find an alternative way. Our proposed methodology uses a machine-learning algorithm called Linear Regression to uncover alternative network paths. To provide for speedy failure recovery, the controller communicates this alternate path to the network switches ahead of time. We train, test, and validate the learning model using a machine learning approach. To simulate our proposed technique and locate the trials, we use the Mini net network simulator. The simulation results show that our suggested approach recovers link failure most effectively compared to existing solutions.