Network Traffic Prediction Algorithm Based On Improved Support Vector Machine

The Internet is the crystallization of human wisdom, the major scientific and technological inventions of the20th century, and an important symbol of contemporary advanced productive forces. The Internet has a profound impact on the world’s economic, political, cultural and social development. It promotes the transformation of social production and living and dissemination of information. In China, there is a growing popularity of the Internet, and then a growing need of network service. Original business like search engines, online advertising, online news, online games, instant messaging are developing rapidly. Videos on line, data offline download are emerging. Meanwhile, the numbers of Internet users are dramatic increasing, making a increasingly complex network behavior. BT download or network attacks often makes networks crush. All these things require us to have a better knowledge of t the network performance of network services such as delay, bandwidth management, an in-depth understanding of health and behavioral characteristics of network.The rapid development of Internet makes a high requirement on network management. Network traffic prediction is the key issue of network management. Network traffic prediction accuracy, timing, is directly related to the efficiency and performance of the network management. It is due to the important position of network traffic prediction, more and more researchers turn into this research field.This article focuses on the network traffic prediction algorithm based on improved support vector machine, not only offers some theoretical analysis, but also offers an instance validation with the MATLAB platform. The results show that our algorithm is more accuracy than the BP neural network algorithm in the network perdition.The main results of this paper are as follows:(1) Offering an analysis of network flow prediction theory, including the linearity and nonlinearity treats of the network traffic, and also the theoretical basis of network traffic prediction and forecast conditions. Accuracy and other features.(2)Summarize the research progress of the network traffic prediction algorithm.offers a analysis of a variety of different present prediction algorithm models, pointing out where the problem lies, and the propose a new network traffic prediction algorithm based on improved support vector machine.(3) Due to the parameter optimization problems of the for the support vector, we introduce the genetic algorithm to achieve an improved support vector machine model. This enables it to determine the various parameters more quickly and more accurately. Then, this improved support vector machine algorithm was introduced to the network traffic prediction area. With the real network traffic data, we establish an instance test. Compared with the BP neural network algorithm, the improved support vector machine algorithm shows certain advantages in performance.