Abstract: In view of mode mixing caused by EMD(Empirical Mode Decomposition)，this paper proposed a self-similari- ty traffic prediction method based on the combination models. Through the EEMD(Ensemble Empirical Mode Decompo- sition) process, the long-term dependence existing in network traffic was removed effectively. Additionally, according to the different characteristics of each IMF(Intrinsic Mode Function) produced by EEMI)，ANN C Artificial Neural Net- work) and ARMA(Auto Regressive Moving Average) were adopted for different IMFs. The simulation results demon- strate that the proposed method can effectively predict the traffic and has high precision.

Key words: Combination model, Traffic prediction, Ensemble empirical mode decomposition, Intrinsic mode function