Routing Enhancement Technology Research Based On Machine Learning

With the advent of the era of big data,deep learning technology has become a research hotspot in the field of artificial intelligence.Deep learning has become a breakthrough machine learning technology in various fields in computer science and other disciplines.It has been in image recognition and speech recognition.The fields of natural language processing,search recommendation,etc.have shown great advantages and continue to evolve.However,the application of deep learning in network traffic control in the network is a relatively new field.With the development of wireless networks,with the development of networks,effective network traffic control such as routing methods in the network has become a key challenge.This is because traditional routing protocols have not learned from their previous experience with network anomalies(such as congestion,etc.).Therefore,intelligent network flow control methods are critical to avoid this problem.In this article,we will implement routing through artificial intelligence to learn historical network status to predict congestion in the network.The main contents of this article are as follows:The advantages and disadvantages of the existing traditional routing protocol OSPF are analyzed.The artificial intelligence method is used to route the OSPF based on the shortcomings of the protocol.The design uses the convolutional neural network to carry out deep learning.In order to make full use of the characteristics of the receptive field in the convolutional neural network,a specific network description is designed as input,and the state description and flow distribution description of the router are designed.Suitable for input as a convolutional neural network.3.The structure of the convolutional neural network is designed.The specific convolutional layer and pooling layer design are completed for the input,and the specific activation function is selected to ensure the gradient loss.Then,the functions of data collection and real-time routing of routers in the network are designed through centralized control.In order to avoid the impact of insufficient training data,a real-time training method is designed to train the convolutional neural network in real time through the network operation data to achieve better results.4.By simulating the network state in some scenarios and comparing with thetraditional OSPF protocol,the ideal results are obtained,which proves that artificial intelligence can be used in the case of burst traffic and centralized traffic at certain points.Low average latency and lower packet loss rate.And when the load is higher,it also has better performance than the OSPF protocol.The design of artificial intelligence routing has a greater advantage over traditional network routing protocols in terms of congestion.Due to the advantages brought by real-time training,the artificial intelligence routing effect is better with the running time.Thus learning from the historical congestion.