Queue response to input correlation functions: continuous spectral analysis
IEEE/ACM Transactions on Networking (TON)
IEEE/ACM Transactions on Networking (TON)
Timescale of interest in traffic measurement for link bandwidth allocation design
INFOCOM'96 Proceedings of the Fifteenth annual joint conference of the IEEE computer and communications societies conference on The conference on computer communications - Volume 2
Broadband traffic modeling: simple solutions to hard problems
IEEE Communications Magazine
Fast algorithms for measurement-based traffic modeling
IEEE Journal on Selected Areas in Communications
| Hi-index | 0.01 |
The CMPP (Circulant Modulated Poisson Process) modeling approach represents an appealing solution since it provides the integration of traffic measurement and modeling. At the same time, it maintains the Markovian hypothesis that permits analytical transient and steady-state analyses of queueing systems using efficient algorithms. These relevant features of CMPP approach has driven us to analyze in more details the fitting procedure when it is applied to actual broadband traffic. In the paper, investigating the estimation algorithm of model parameters, we emphasize the difficulty of CMPP in capturing the upper tail of marginal distribution of actual data, which leads to an optimistic evaluation of network performance. As shown in the paper, a simple relation exists between the number of significant eigenvalues obtained by the spectral decomposition and the peak rate that the CMPP structure is able to capture. The relation evidences the difficulties of CMPP to model actual traffic, characterized by long tailed distribution, as well as traffic data with the well accepted hypothesis of gaussian marginal.