Impact of Short-range Dependence on the Performance of Network Analysis Methods

Lin Qingjia,Chen Di,Liu Yuncai
DOI: https://doi.org/10.13382/j.jemi.2006.05.015
2006-01-01
JOURNAL OF ELECTRONIC MEASUREMENT AND INSTRUMENT
Abstract:LRD (long-range dependence) and SRD (short-range dependence) behaviors of traffic influence network modeling, service providing and traffic engineering. In order to understand this influence, several network analysis methods are applied to two kinds of simulated traces. One trace is generated by FGN model with LRD structure, another one is generated by FARIMA model with LRD and SRD simultaneously. Their performances are compared and they arrived at the conclusions: These methods exhibit good performance under the conditions with LRD structure, but they exhibit poor performance under that with LRD and SRD structure simultaneously. In order to reduce the impact of SRD, an aggregated method is presented. The results show that this method can eliminate or at least reduce the SRD structure. So it can be used as a tool for analyzing real traffic trace.
What problem does this paper attempt to address?