|
Addis Ababa University Libraries Electronic Thesis and Dissertations: AAU-ETD! >
Faculty of Technology >
Thesis - Electrical Engineering >
Please use this identifier to cite or link to this item:
http://hdl.handle.net/123456789/3799
|
| Title: | INVESTIGATION OF ETHERNET TRAFFIC |
| Authors: | Yalemzewd, Negash |
| Advisors: | Prof. Devarajan |
| Copyright: | Jun-2001 |
| Date Added: | 19-Nov-2012 |
| Abstract: | A common assumption in modeling computer networks is that packet arrivals occur as
a Poisson process. However, data communication traffic levels fluctuate over time,
and delays through congestion can occur even on lightly utilized links. These
fluctuations can occur over very short periods of time giving rise to the concept of a
burst of traffic. Bursts of traffic can be of intensity more than five times the average
utilization so that if a user is trying to send data and it coincides with a burst the user
will experience delays. Traffic that exhibits these wild fluctuations is known as
"bursty" traffic.
Understanding the nature of traffic in high-speed, high-bandwidth communications
systems such as B-ISDN is essential for engineering, operations, and performance
evaluation of these networks. In a first step towards this goal, it is important to know
the traffic behavior of some of the expected major contributors to future high-speed
network traffic. In this research work traffic from an Ethernet LAN is studied for it
behavior.
The purpose of this work is to show the self-similarity nature and long-range
dependence of Ethernet network traffic. Different mathematical and graphical
techniques are used to show this behavior. The result indeed shows the long-range
dependence or the presence of long memory in Ethernet data traffic. A graphical proof
of the self-similarity nature of the traffic is shown. Also Fractional Auto-Regressive
Integrated Moving Average (FARIMA) model is developed to capture the long as
well as the short memory properties of the collected Ethernet traffic data. The model
is found to be in good agreement with the periodogram calculated from the data. The
model could be used in different network application like congestion control in highbandwidth
networks, bandwidth allocation and the like. All the results in this work are
supported by a rigorous statistical analysis of the collected data coupled with a
discussion of the underlying mathematical and statistical properties of long memory
processes. |
| URI: | http://hdl.handle.net/123456789/3799 |
| Appears in: | Thesis - Electrical Engineering
|
Items in the AAUL Digital Library are protected by copyright, with all rights reserved, unless otherwise indicated.
|
