Cloud traffic prediction based on fuzzy ARIMA model with low dependence on historical data

Abstract Traffic prediction with high accuracy has become a vital and challenging issue for resource management in cloud computing. It should be noted that one of the prominent factors in resource management is accurate traffic prediction based on a few data points and within a short time period. The autoregressive integrated moving average (ARIMA) model is a suitable model to predict traffic in short time periods. However, it requires a massive amount of historical data to achieve accurate results. On the other hand, the fuzzy regression model is adequate for prediction using less historical data. Aforementioned by these considerations, in this paper, a combination of ARIMA and fuzzy regression called fuzzy autoregressive integrated moving average (FARIMA) is used to forecast traffic in cloud computing. Besides, we adopt the FARIMA model by using the sliding window, called SOFA, concept to determine models with higher prediction accuracy. Accuracy comparison of these models based on the root means square error and coefficient of determination demonstrates that SOFA is about 5.4 and 0.009, respectively, which is the superior model for traffic prediction.