Design And Implementation Of Network Traffic Prediction System Based On The Gray Theory And Least Squares Support Vector Machine

With the rapid development of Internet, the increasing complexity of networkbehavior characteristics have brought great challenges to network planning and networkmanagement.However,contradictions have become increasingly prominent, networktraffic analysis and the establishment of an effective network traffic model is the hotissues of the current network research.The research and analysis of network trafficcharacteristics and in-depth understanding of the inherent nature of the network, and tounderstand the situation of fundamental methods and means of operation of the networkperformance, optimize network design and implementation of the important way oftraffic engineering.Based on the gray theory and least squares support vector machine(LSSVM), to create a new network traffic prediction model. Based on the actualnetwork traffic data, for instance analysis, the results proved that the new model andalgorithm has high accuracy and universality. For network traffic forecast, do thefollowing work:(1) on the basis of in-depth study of gray GM (1,1) prediction model and itsimproved method that affect the predictive accuracy of the model, and considered theaffect of the old and new data for prediction, this paper proposes a weighted method toprocess the raw data, and a more reasonable and more simple method that thereconstructed model background value and an estimate of the initial value.(2) on the basis of in-depth study on support vector machine, based on thecharacteristics of the network traffic data, the paper proposes a dynamic sound LSSVMmodel. In order to verify the correctness and validity of the algorithm of robust LSSVM,general LSSVM model and robust LSSVM model simulation experiments.(3) based on the combination of gray model with least squares support vectormachine, and the model of the characteristics of the data processing, the paper will graysquares support vector machine is divided into three categories: parallel, series andresidual type. Experimental results show that the combination forecasting model caneffectively improve the accuracy of the predicted higher fitting accuracy than a singlemodel, and has a certain value. (4) In this thesis, gray squares support vector machine prediction model for themain structure of the main structure of the network traffic prediction system design, andcan achieve all calculations.The major functional blocks of the network trafficprediction system can be divided into three: the flow of data processing module, trafficprediction module and flow diagram generation module.This system acquires a bestpredicting model through much experiment; by means of this system, an effectiveprediction can be made on the network flow in the short-term period and on theinformation, which offers an active and beneficial exploration in the network planningand network management.