The Research Of Combined Prediction Model Of Network Traffic With Grey Theory And Neural Network

Predicting and modeling network traffic is always an important subject in network capability studying. The prediction results can bring out essential reference for bandwidth allocation, network traffic control, routing control, entry control and error control in network management. Network traffic is dynamical, realtime, correlative, stochastic and noisy. The forecasting precision and expressive power of the model is very important for analysis, simulation and network behavior study.In this paper, network traffic prediction is the study background. By comparing several collection tools, we select Cacti to build our collection system and collect data from network switching in our campus. After analyzing, we found network traffic is time series. We built GM(1,1) model and equal dimension Grey model. And we build Verhulst model for Internet access quantity, which shows saturated "S" shape. Some tools were used to model and predict and the calculate results were analysed.This paper advanced a kind of grey predicting method based on wavelets and give out a new combined prediction model with grey theory and neural network (adopting BP network, LM algorithm, and radial redical network). Such methods were used to predict and making analysis for the data we selected. Precision and effect of the model has also been analysed. Foresaid work laid the foundation for developing short term (and long range) network traffic prediction software system.Aiming at the shortages of common dynamic integrate linking status method and network throughput based routing algorithm, a new optimized routing method was brought out. It can well complement some existent algorithm. Theoretical analytic result shows, routing effect is better.At the end of the paper, the author summarized the study's application area in internet routing decision, virus detecting in Local Area Network, prevention illegal invasion and commercial business decision.