Title :
Research on sampling collecting and predicting for IP network traffic
Author_Institution :
Dept. of Comput. Sci. & Technol., Hunan Inst. of Technol., Hengyang, China
Abstract :
Measurement and prediction of network traffic is the base of network management and performance analysis. In this paper, a sampling algorithm based on hash temporary and mask match was put forward, and estimating actual traffic method from sampling data was given. Experiment results show that max estimation error is only 8.26%. Then, by training experiments, neuron number of input layer and hidden layer was identified and a 5*4*3 BP neural network model was set up, BP algorithm was improved used adaptive learning rate. Experiment results validated the correctness and accuracy of the BP neural network model, and proved the prediction precision was higher than that of grey model.
Keywords :
IP networks; backpropagation; computer network management; computer network performance evaluation; learning (artificial intelligence); neural nets; sampling methods; telecommunication traffic; BP neural network model; IP network traffic; adaptive learning rate; grey model; hash temporary; mask match; network management; neuron number; performance analysis; sampling algorithm; sampling collecting research; training experiments; Artificial neural networks; Frequency modulation; IP networks; Planning; Reactive power; BP Neural Network; Network Traffic; Sampling Collecting; Traffic Predicting;
Conference_Titel :
Computer Science and Network Technology (ICCSNT), 2011 International Conference on
Conference_Location :
Harbin
Print_ISBN :
978-1-4577-1586-0
DOI :
10.1109/ICCSNT.2011.6182216
