Skip to Main Content
Network providers are often interested in providing dynamically provisioned bandwidth to customers based on periodically measured nonstationary traffic while meeting service level agreements (SLAs). In this paper, we propose a dynamic bandwidth provisioning framework for such a situation. In order to have a good sense of nonstationary periodically measured traffic data, measurements were first collected over a period of three weeks excluding the weekends in three different months from an Internet access link. To characterize the traffic data rate dynamics of these data sets, we develop a seasonal autoregressive conditional heteroskedasticity (ARCH) based model with the innovation process (disturbances) generalized to the class of heavy-tailed distributions. We observed a strong empirical evidence for the proposed model. Based on the ARCH-model, we present a probability-hop forecasting algorithm, an augmented forecast mechanism using the confidence-bounds of the mean forecast value from the conditional forecast distribution. For bandwidth estimation, we present different bandwidth provisioning schemes that allocate or deallocate the bandwidth based on the traffic forecast generated by our forecasting algorithm. These provisioning schemes are developed to allow trade off between the underprovisioning and the utilization, while addressing the overhead cost of updating bandwidth. Based on extensive studies with three different data sets, we have found that our approach provides a robust dynamic bandwidth provisioning framework for real-world periodically measured nonstationary traffic.