Research On Network Traffic Modeling Analysis And Prediction Based On Particle Swarm Optimization FARIMA Model

As 5G technology gradually spreads,the network environment becomes more and more complex,resulting in network traffic congestion,attacks,packet loss and other problems,and effective prediction of network traffic can help to complete the protocol design,traffic scheduling,network attack detection,etc.Therefore,the modeling and prediction analysis of network traffic becomes the key to solve network problems.To improve the accuracy of network traffic prediction,the FARIMA model(Fractional Difference Autoregressive Sliding Average Model),which can portray both long and short correlations,is chosen to predict the network traffic.In order to reduce the complexity of the FARIMA model calculation,the FARIMA model is decomposed into two parts: the fractional difference and the ARMA model.Meanwhile,in order to improve the prediction effect of FARIMA model,particle swarm-genetic hybrid algorithm is introduced to optimize the parameters of FARIMA model.By testing function,compare the superiority-seeking ability of particle swarm algorithm,genetic algorithm,and hybrid particle swarm-genetic algorithm.The optimized FARIMA model was validated by applying the optimized FARIMA model to real flow forecasting in Bellcore Labs,and the results showed that the optimized FARIMA model provided better forecasting results than the traditional FARIMA model.To further reduce the prediction error of the FARIMA model optimized by the hybrid particle swarm-genetic algorithm,the IGM model(Improved Gray Prediction Model)is introduced to enhance the prediction effect of the FARIMA model.A correction factor is introduced to the differential equation of the traditional gray prediction model to form the IGM model,thus increasing the stability of the traditional gray prediction model.For the volatility of network traffic,the FIGARCH model is used to portray,and the FIGARCH model is combined with the FARIMA model(FARIMA-FIGARCH model),i.e.,the long correlation and volatility of network traffic are portrayed simultaneously.The prediction error of the FARIMA-FIGARCH model is further reduced using the improved gray prediction model.By comparing the above two improved FARIMA models,the FARIMA-FIGARCH model improved by the IGM model can significantly improve the accuracy of network traffic prediction,and the model can be widely used in financial time series prediction,engineering deformation detection and other fields.