Study Of The KEY Technologies Of Soft Computing-based IP Network Traffic Monitoring Prediction And Flow Control

Accompanied by the rapid development of Internet technology and network applications,the Internet has become an indispensable part of scientific research and people’s daily life. Itis extremely important to transmit data rapidly and effectively in the Internet for the real-timeservice, such as multimedia video and VoIP, as well as the transmission of non real-timeservice like file-downloading. Therefore, achieving the optimal use with minimum cost of thenetwork resource is one of the focus research areas today. IP network monitoring and controltechnology provides an effective way to transmit data in a limited network bandwidthenvironment with better services. Combining with soft computing technique, this thesisstudies the monitoring model, traffic prediction algorithms and flow control method of theInternet：(1) An Internet QoS monitoring model based on agent technology is proposed in thisthesis and the structure and function of the model is described. Based on it, the componentdesign and implementation of the model is also discussed. Being as the research object of themodel, the QoS of VoIP services is studied and the algorithm of the call detail records isoptimized.(2) Combined with the advantages of gene expression programming, an ITF-EBPalgorithm (Internet Traffic Forecasting Based on Evolutionary BP Neural Network) is putforwarded in the thesis which is based on BP neural network-based IP network trafficprediction algorithm. The thesis describes the encoding, the genetic manipulation and thealgorithm exhaustively. Compared with the traditional algorithm, ITF-EBP could improve thetraining speed and global search capability. Simulation and performance analysis prove it.The thesis describes the encoding, the genetic manipulation and the algorithm exhaustively,and compares the simulation and performance analysis with the traditional algorithm.(3) Based on studies on the wavelet transform and FARIMA technology, a networktraffic prediction algorithm based on wavelet transform and modified GFARIMA model isproposed in the thesis. This algorithm decomposes the original flow after its treatment,reconstructs the detail component and the approximate component respectively, predicts each component via modified GFARIMA prediction algorithm, and synthesis the flow. It improvesthe accuracy of the prediction algorithm and partly eliminates the lagging effect of theoriginal method.(4) An ORS-SAGEP (Optimization for Route Selection based on Simulated AnnealingGene Expression Programming) algorithm based on ORS-GEP (Optimization for RouteSelection based on Gene Expression Programming) algorithm is brought forward in thisthesis. As for the advantages of easy jumping out from local optimal solution of the geneexpression programming, ORS-SAGEP could improve the programming accuracy, and thesimulation results prove its performance.In conclusion, thesis provides a whole set of IP network traffic monitoring and controlsolution, it describes the establishment of the monitoring model, the network trafficprediction algorithm and the QoS flow control. The solution has been verified the feasibility,and it can play an objective inspiration in actual scene...