Adaptive bandwidth provisioning envelope based on discrete temporal network measurements

We propose an adaptive bandwidth provisioning mechanism based on packet-level measurements (done at discrete time intervals) without any assumption on the statistical property of the traffic. Our approach consists of two stages: traffic forecasting, followed by bandwidth estimation. For the first stage, we have developed a probability-hop forecasting algorithm (based on ARIMA model) to forecast traffic based on online traffic measurements in a temporal management information base framework. For the second stage, we present several bandwidth provisioning schemes that allocate or deallocate bandwidth depending on the traffic forecast generated by our forecasting algorithm. We have introduced an utility function (along with other measures) to evaluate the overall effectiveness of our bandwidth provisioning framework. Through the use of real-world data, we have found that our approach works well for highly fluctuating data