ANN based NGN IP traffic prediction in Thailand

This paper presents a study of using artificial neural network (ANN) in predicting IP traffic fluctuation in IP based next generation network (NGN) in Thailand. By analyzing the time level correlation of traffic flow over TOT IP core network, monitor by the Multi Router Traffic Grapher (MRTG) from January,2006 to December, 2006 at Latya and Phrakhanong exchange. For comparison purposes, the back propagation learning algorithms are considered, in particular ,optimization techniques with statistical methods prediction such us Double Exponential smoothing, Lest Square Methods and Adaptive response rate simple exponential smoothing method are applied for predicting the same node of problems. Computational results by using the data monitoring at Latya exchange for training and actual traffic in Phrakhanong area to check prediction correctness show that the based ANN method obtained better results, it can be prediction more precisely.