Traffic Prediction Method for Broadband Satellite Communication Systems

Considering the significant burstiness in self-similar network traffic, this paper presents a one-step traffic prediction method based on data decomposition. Firstly control chart is used to decompose the original traffic into non-burst part and burst part. Then the two parts are predicted by ANN and adaptive template matching method, respectively. Finally, the prediction of the original traffic is obtained by fusing the predictions of the two parts. Simulation results show that the accuracy of method proposed in this paper is much higher compared with the traditional traffic prediction methods.