A Hybrid Prediction for Non-Gaussian Self-Similar Traffic

摘要

There is growing evidence shows that non-Gaussian, namely heavy tailness is the key cause of burstiness in self-similar traffic. We present three predictors including autoregressive (AR), moving average (MA) and fractional autoregressive integrated moving average (FARIMA) based on the symmetrical non-Gaussian self-similar traffic model. The three predictors can minimize the dispersion according to the minimum dispersion criteria with infinite variance. The final predicted values are attained by combining the previous three individual predicted values. Our predicted results for the actual trace collected from Bellcore Lab and Lawrence Berkeley Lab show that the three individual predictors are precise and reliable, the compound predictors can enhance the final predicted accuracy.