Diffusion wavelet-based analysis on traffic matrices by different diffusion operators

Hui Tian, Binze Zhong, Hong Shen

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)


Traffic matrix (TM) describes the traffic volumes traversing a network from the input nodes to the output nodes over a measured period. Such a TM contains very useful information for network managers, traffic engineers and users. However, TM is hard to be obtained and analyzed due to its large size, especially for large-scale networks. In this paper, we present a new method based on diffusion wavelets for analyzing the traffic matrix. It is shown that this method can conduct efficient multi-resolution analysis (MRA) on TM. We compare the analysis results by using different diffusion operators. Through reconstructing the original TM from the diffused traffic on a particular level, we show the high efficiency of this MRA tool based on these operators. We then develop an anomaly detection method based on the analysis results and explore the possibilities of other potential applications.

Original languageEnglish
Pages (from-to)1874-1882
Number of pages9
JournalComputers and Electrical Engineering
Issue number6
Publication statusPublished - Aug 2014
Externally publishedYes


Dive into the research topics of 'Diffusion wavelet-based analysis on traffic matrices by different diffusion operators'. Together they form a unique fingerprint.

Cite this