A forecasting model of communication traffic based on false nearest neighbor

Forecast of communication traffic is an important part of the communications network management. In view of the communication traffic feature, based on the conception of false nearest neighbor in chaos theory, the best embedded dimensionality and time delay are determined simultaneously, the nonlinear time series are reconstructed and the sensibility for change of communication traffic is improved. So the input vector space of forecast model is simplified. Then the forecast model is obtained by RBFNN regression modeling. By comparing simulation results, the prediction model based on FNN-RBFNN has simpler construction, higher precision and more universal than those of the forecasting model based on RBFNN. The experimental results show that the method is very effective.