Abstract:

The RTMFP protocol of FlashP2P technique is analyzed in detail and a FlashP2P traffic recognition algorithm based on RTMFP packet detection is proposed.The algorithm is used to effectively identify the FlashP2P streams of the domestic mainstream video sites. On this basis,the characteristics of FlashP2P flow are analyzed and their selfsimilarity is proved.Finally, based on the empirical mode decomposition of the ARIMA, we build a model to predict the selfsimilar network traffic. The simulation results show that the model not only reduces the complexity of the algorithm but also has high shortterm forecast accuracy.

Key words: FlashP2P;traffic identification;EMD;traffic forecasting