Bandwidth variability prediction with rolling interval least squares (RILS)
Proceedings of the 50th Annual Southeast Regional Conference
Improving network measurement efficiency through multiadaptive sampling
TMA'12 Proceedings of the 4th international conference on Traffic Monitoring and Analysis
A multiadaptive sampling technique for cost-effective network measurements
Computer Networks: The International Journal of Computer and Telecommunications Networking
| Hi-index | 0.00 |
In a internet architecture called CABO, multiple routing architectures can run on a shared physical infrastructure, which is carried out with network virtualization. This paper presents the design and evaluation of a bandwidth allocation algorithm based on multi-commodity flow problem solver integrated with a traffic predictor: linear predictor with dynamic error compensation (L-PREDEC). The basic idea of our design is that some failure in the MFP computation implies that one or more links do not have enough available capacity, which violates the linear constraints on the commodities for each link when modeling MFP. To avoid producing bottleneck links, we employed traffic predictor. On one hand, MFP solver makes better resource utilization by making use of the thin pieces of available bandwidth, by which the virtual network can accept more service requests. On the other hand, the traffic predictor adjusts the link with the largest occupation (bottleneck link) by periodically monitoring the traffic rate of a user link and adjusting the reserved bandwidth based on the prediction made from the traffic history. Then we present the results of performance comparisons of the predictor-integrated algorithm and the allocation algorithm only by Solving MFP. The comparisons are based on the mean packet delay, the variance of the packet delay, and the buffer requirements. Our performance tests show that predictor-integrated algorithm works better than the allocation algorithm only by Solving MFP in terms of the three metrics listed above.