We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.NI

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Networking and Internet Architecture

Title: Time Series Analysis of Computer Network Traffic in a Dedicated Link Aggregation

Abstract: Fractal behavior and long-range dependence are widely observed in measurements and characterization of traffic flow in high-speed computer networks of different technologies and coverage levels. This paper presents the results obtained when applying fractal analysis techniques on a time series obtained from traffic captures coming from an application server connected to the Internet through a high-speed link. The results obtained show that traffic flow in the dedicated high-speed network link have fractal behavior when the Hurst exponent is in the range of 0.5, 1, the fractal dimension between 1, 1.5, and the correlation coefficient between -0.5, 0. Based on these results, it is ideal to characterize both the singularities of the traffic and its impulsiveness during a fractal analysis of temporal scales. Finally, based on the results of the time series analyses, the fact that the traffic flows of current computer networks exhibit fractal behavior with a long-range dependency is reaffirmed.
Comments: 7 pages, 2 figures, 6 tables, preprint for IEEE Transactions on Information Theory 2021
Subjects: Networking and Internet Architecture (cs.NI); Signal Processing (eess.SP); Physics and Society (physics.soc-ph)
MSC classes: 68Q11 (Primary), 94A12 (Secondary)
ACM classes: F.2.1; G.2.0
Cite as: arXiv:2107.05484 [cs.NI]
  (or arXiv:2107.05484v1 [cs.NI] for this version)

Submission history

From: Ginno Millán [view email]
[v1] Tue, 11 May 2021 22:41:05 GMT (330kb)

Link back to: arXiv, form interface, contact.