Content uploaded by Khaled Elleithy
Author content
All content in this area was uploaded by Khaled Elleithy
Content may be subject to copyright.
A Novel Simulation Based Methodology for the
Congestion Control in ATM Networks
Heven Patel, Hardik Patel, Aditya Jagannath, Syed Rizvi, Khaled Elleithy
Computer Science and Engineering Department
University of Bridgeport, Bridgeport, CT
Abstract- In this project, we use the OPNET simulation tool
for modeling and analysis of packet data networks. Our project
is mainly focused on the performance analysis of
Asynchronous transfer mode (ATM) networks. Specifically, in
this project, we simulate two types of high-performance
networks namely, Fiber Distributed Data Interface (FDDI)
and Asynchronous Transfer Mode (ATM). In the first type of
network, we examine the performance of the FDDI protocol by
varying network parameters in two network configurations. In
the second type, we build a simple ATM network model and
measure its performance under various ATM service
categories. Finally, we develop an OPNET process model for
leaky bucket congestion control algorithm and examine its
performance and its relative effect on the traffic patterns (loss
and burst size) in an ATM network. Our simulation results
show that the ATM network has longer response time than
FDDI. On the other hand, it shows that for both token ring
and MAC delay, ATM is shorter than FDDI
I. INTRODUCTION
ATM is the new generation of computer and
communication networks that is being deployed
throughout the telecommunication industry as well as in
campus backbones. ATM technology distinguishes itself
from the previous networking protocols in that it has the
latest traffic management technology and thus allows
guaranteeing delay, throughput, and other performance
measures. This report describes key features of the ATM
network and some relative simulation work we have done
in OPNET. FDDI and ATM are two well known
technologies used in today’s high-performance packet
data networks. We use OPNET to simulate networks
employing these two technologies. FDDI network is an
older and well-established technology used in LAN’s.
FDDI is a networking technology that supports 100Mbps
transmission rate, for up to 500 communicating stations
configured in a ring or a hub topology. In a dual ring
topology, maximum distance is 100 km. FDDI supports
three types of devices: single attachment stations, dual-
attachment stations and this topology provide high degree
of fault tolerance. Since OPNET does not support dual-
attachment stations, we used scenarios with single –
attachment stations connected in a hub topology with
FDDI concentrators. Our simulation scenarios include
client server and source destination networks with various
protocol parameters and service categories. We also
simulate a policing mechanism for ATM networks .In this
section we simulate the performance of the FDDI
protocol. We consider network throughput, link
utilizations, and end-to-end delay by varying network
parameters in two network configurations. The leaky
bucket mechanism limits the difference between the
negotiated mean cell rate (MCR) parameter and the actual
cell rate of a traffic source. It can be viewed as a bucket,
placed immediately after each source. Each cell generated
by the traffic source carries token and attempts to place it
in the bucket. If the bucket is empty, the token is placed
and the cell is sent to the network. If the bucket is full, the
cell is discarded. The size of the bucket is equal to an
upper bound of the burst length, and it determines the
maximum number of cells that can be sent consecutively
into the network.
A. Asynchronous Transfer Mode (ATM)
In this section we present the OPNET
implementation of the leaky bucket congestion control
algorithm. In ATM networks, channels do not have fixed
bandwidths. Thus, users can cause congestion in the
network by exceeding their negotiated bandwidth.
Prohibiting users from doing so (policing) is important,
because if excessive data enters the public ATM network
without being controlled, the network may be overloaded
and may encounter an unexpected high cell loss. This cell
loss affects not only the violating connections, but also
the other connections in the network. This degrades the
network functionality. As shown in Figure-1 we create a
scenario with five different clients connected by central
hub and that hub in turn connected to a main server
through 10 base-T links. Then the simulation is run and
compared with the result of FDDI. The average of Ftp
response time in second is observed. From Figure-2 we
concluded that ATM gives higher response time then
FDDI.
B. Fiber Distributed Data Interface (FDDI)
Our next main goal to simulate the same scenario as
we did on ATM. We create two different scenarios for
FDDI. First as shown in Figure-3 is a ring topology
configuration. In this configuration, the network in
connected in a ring fashion. Further each hub is connected
to the station through a 10 Base T connection.Figure-4
shows a client server configuration connected via two
switches with 100Base-T link. The delay (sec) and
throughput (sec) are observed from the first and second
configurations respectively.
C. Leaky Bucket Algorithm
The leaky bucket mechanism limits the difference
between the negotiated MCR parameter and the actual
cell rate of a traffic source. It can be thought of as a
bucket placed immediately after each source. The traffic
source generates cells. Each cell thus generated carries a
token and attempts to place it in the bucket. If the bucket
is empty, the token is placed and the cell is sent to the
network. If the bucket is full, the cell is discarded. The
bucket gets emptied at a constant rate equal to the
negotiated MCR parameter of the source.
The size of the bucket is equal to an upper bound of
the burst length, and it determines the maximum number
of cells that can be sent consecutively into the network.
The model is shown in the above figure. From the figure
we observe that the process can reach the arrival state
when the packet arrives or can reach an idle state where it
waits for the packets. This can happen starting from the
initial state. Depending on whether the bucket is full or
empty, from arrival state, the process reaches either serve
or drop state. Users can change the following parameters
in the leaky bucket process model leaking rate bucket
size.
Figure-2 average response time(sec)
Figure-
1 ATM with servers, hub and switches network in
client-server application
Figure.3 Ring topology configuration
Figure.4 Client-server configuration
D. Configuration Parameters
Token Ring Station Latency (fddi_tr_slip8_gtwy): 4
bits Switch BPDU Service Rate (fddi16_switch): 500,000
packets per second. Switch packet Switching Speed
(fddi16_switch): 500,000 pkts/sec. ATM Switching speed
(atm4_crossconn): infinity.
IP Forwarding Rate (subnet router): 50,000
packets/seconds. IP Ping traffic (subnet router): None. We
set applications running as: Email (heavy), File Transfer
(heavy), Telnet Session (heavy), Web Browsing (heavy).
To do this, In Profile configuration: Start time:
Exponentially Distributed, Mean Outcome 100 seconds
Start time offset Exponentially Distributed, Mean
Outcome 10 seconds. Operation mode: Simultaneously
Duration: end of Profile.
Some workstations in susbnet1 and 2 are clients of the
applications supplied by the four servers on the subnet3.
We ran simulation for 60 minutes, and collected statistics
such as service Response Time, Token Ring Delay and
Token Ring MAC Delay.
II. CONCLUSION AND FUTURE WORK
In this paper we focused on simulating two commonly
used packet data network technologies: FDDI and ATM.
We simulated two FDDI and one ATM network
scenarios. From that we concluded that ATM has longer
response time than FDDI, while for the token ring delay
and MAC delay, ATM is shorter than FDDI. Also this
paper came out with a solution to the congestion control
by implementing the famous leaky bucket algorithm in
OPNET.
REFERENCES
[1] Available at:
http://www.cs.wustl.edu/~jain/cis78895/ftp/atm_cong/index.
html
Figure-5 Average ring token delay (sec)
Figure-6 Average point to point throughput (bits/sec)
Figure-7 Architecture of Leaky Bucket Algorithm
Figure-8 average token access delay(sec)
[2] ANSI X3.139-1987: Fiber Distributed Data Interface (FDDI)
- Token Ring Media Access Control (MAC)
http://www.ansi.org.
[3] ATM Forum. Traffic Management SpecificationVersion4.0.
ftp://ftp.atmforum.com/pub/approved-specs/aftm
0056.000.pdf
[4] G.Niestegge,“The leaky bucket policing method in
asynchronous transfer mode networks,” International
Journal of Digital and Analog Communication Systems, vol.
3, pp. 187-197, 1990.
[5] OPNET Contributed Model Depot:
http://www.opnet.com/services/depot/home.html.
