A new scheme for traffic estimation and resource allocation for bandwidth brokers
Computer Networks: The International Journal of Computer and Telecommunications Networking
Dynamic allocation of resources to virtual path agents
IEEE/ACM Transactions on Networking (TON)
Filtering and forecasting problems for aggregate traffic in internet links
Performance Evaluation
Supporting real time VBR video using dynamic reservation based on linear prediction
INFOCOM'96 Proceedings of the Fifteenth annual joint conference of the IEEE computer and communications societies conference on The conference on computer communications - Volume 3
An adaptive virtual path allocation policy for broadband networks
INFOCOM'96 Proceedings of the Fifteenth annual joint conference of the IEEE computer and communications societies conference on The conference on computer communications - Volume 1
| Hi-index | 0.24 |
Managing the bandwidth allocated to a Label Switched Path in MPLS networks plays a major role for provisioning of Quality of Service and efficient use of resources. In doing so, two main contrasting factors have to be considered: not only the bandwidth should be adapted to the traffic profile but also the effort for bandwidth renegotiation associated with a variation of the allocated bandwidth should be kept at low levels. In this context, we formulate a problem of optimal LSP bandwidth reservation as the one of minimizing a convex combination of the difference between the assigned bandwidth and the estimated future traffic, and of a measure of the frequency of bandwidth variations. The contribution of this paper is to propose a new method to reserve optimally the bandwidth of an LSP, avoiding an excess of bandwidth renegotiations on the basis of prediction of future traffic, assuming a simple birth-and-death model to describe the traffic dynamics. Whenever the prediction is inaccurate due to unpredictable variations in the characteristics of real traffic, a suitable ''emergency procedure'' is proposed, which performs a new traffic prediction and a consequent modified bandwidth reservation. Numerical results are presented which show the effectiveness of the method and the achieved performance, both for simulated and real data traffic.