Direct Research Journal of Engineering and Information Technology Vol.1 (3), pp. 54-63,September 2013
Available online at http://directresearchpublisher.org/drjet © 2013 Direct Research Journals Publisher
Original Research Paper
Dropping policies for transient
network traffic congestion control
Abah Joshua
1
*,Alibaba Dauda
2
and Abah O.
Sunday
3
1
Department of Computer Engineering, Faculty
of Engineering, University of Maiduguri,Nigeria.
2
Computer and Internet Centre, University of
Maiduguri,Nigeria.
3
Field Engineer, Coollink Lagos, Nigeria.
ABSTRACT
This paper examines the effects of dropping policies
on dropping probabilities for multi-class traffic and
compares their performance and support for Quality
of Service (QoS). The flexibility of these dropping
policies is illustrated with respect to performance
parameters and traffic characteristics. The method
used for this evaluation is simulation on a model
network that carries traffic of different class; real-
time and bursty (non real-time) traffic. Results
showed that Weighted Random Early Detection
(WRED) in conjunction with an appropriate queuing
discipline provides better support for Quality of
Service of real time applications while Tail Drop
supports best effort applications.
Key words: Tail Drop, Random Early Detection,
Simulation, Quality of Service, Queuing
*Corresponding Author E-mail: jehoshua_a@yahoo.com,
abah@unimaid.edu.ng
Accepted 22 August, 2013
INTRODUCTION
With rapid growth of the internet, the amount of traffic
(real-time and non real-time) such as voice over IP (VoIP)
calls, multimedia and even non real-time applications like
file transfer protocol (FTP), e-mails etc have been
increased. The requirements of real-time traffic in the
internet are;
i. Low loss rate and
ii. Low end-to-end delay
Routers in the internet have to reflect these requirements
to support real-time traffic.
Active Queue Management (AQM) such as Tail Drop
(DT), Random Early Detection (RED) and Weighted
Random Early Detection (WRED) are usually used to
control congestion in a router and Quality of Service
(QoS) in the entire network (Chaudhuri et al, 2006).
Fortunately, because end-to-end delay is primarily
determined by the buffering, queuing and routing delay of
routers, the same resources are related to end-to-end
delay and congestion control. This means that active
queue management can be used to both control
congestion and support different classes and types of
applications’ traffic in the internet. Among active
management schemes in which dropping policies are
part, this paper seek to compare these schemes to see
which of them has a good feature to bound the queuing
delay and loss rate desired.
MOTIVATION
The digital divide separates the information rich and the
information poor. The Organization for Economic Co-
operation and Development defines the digital divide as
the difference between individuals, households,
businesses and geographic areas with regard to (a) their
opportunities to access ICTs and (b) their use of the
Internet for a wide variety of activities (Emmanuel, 2003).
It is the gap between those who have real access to
information and communications technology and who are
able to use it effectively, and those who don’t have such
access. The digital divide is usually measured in terms of
citizen/population access to ICT. Among the indicators for
measuring access are telephone density (teledensity),
personal computer (PC) deployment and penetration and