Network Traffic Forecast Method Based On GWO-SVR

In recent years,with the continuous development of communication network technology,the 5G era has also come.With the advent of the 5G era,the number of netizens in my country is also increasing,which makes the scale of the network gradually increase to meet people’s increasing network management needs.However,due to the continuous expansion of the network scale,the network traffic has also increased significantly,resulting in uneven network resource allocation and network congestion or overload.Therefore,analyzing and accurately predicting network traffic can effectively manage the network.And maintenance,avoid network congestion,improve network performance,and also play an important role in network security.In view of the diversity and suddenness of today’s network traffic,when using traditional linear models to predict it,the prediction effect will cause large errors.Therefore,in order to improve the accuracy of network traffic prediction,this thesis proposes a multistage prediction method based on gray wolf algorithm and support vector regression.The main work of this thesis is as follows:(1)The related content involved in network traffic prediction is studied,including the characteristics of network traffic and several methods of predicting network traffic at present.And analyzed and summarized the feasibility of using support vector regression algorithm to predict network traffic.(2)Designed and implemented a network traffic prediction method based on gray wolf optimization algorithm and support vector regression algorithm.In this thesis,the gray wolf algorithm is used to optimize the three important parameters in the support vector regression algorithm,and then the SVR prediction model is established to obtain the single-stage prediction model(SGWO-SVR).Then,on the basis of SGWO-SVR,it was first proposed to use a two-stage prediction method to establish a network traffic prediction model,namely TGWO-SVR.(3)Using the public data set of the MAWI working group to verify the prediction method proposed in this article,and compared with the SVR method,GA-SVR prediction method,and DE-SVR prediction method,the model established in this article has an average absolute percentage error and The root mean square error and other performance measurement indicators have good performance,which proves that the prediction model proposed in this thesis has certain advantages.