Global Mobile Information System Simulator in Fedora Linux

Ayyaswamy Kathirvel a , Rengaramanujam Srinivasanb

a Assistant Professor, Faculty of Computer Science and Engineering

b Professor, Faculty of Computer Science and Engineering
EmailID: a [email protected], b [email protected]
B.S.Abdur Rahman University, Chennai – 600 048, Tamilnadu, India.

Abstract: Most network systems are currently build using a layered

Global Mobile Information System Simulator is a popular approach that is similar to the OSI seven layer network
network simulation tool, which is frequently used in the architecture. The plan is to build GloMoSim using a
study of the behavior of large-scale hybrid networks that similar layered approach [4]. Standard APIs will be used
include wireless, wired, and satellite based between the different simulation layers. This will allow
communications are becoming common in both in the rapid integration of models developed at different
military and commercial situations. It is freely available layers by different people [13][14]. It usually made
without fee for education, or research, or to non-profit available on a standalone machine. The goal is to build a
agencies. It is simple to install and use. It is available for library of parallelized models that can be used for the
various Linux flavors files include freebsd-3.3, aix, irix, evaluation of variety of wireless network protocols [5][7].
redhat-6.0, redhat-7.2 and Solaris. This tool is not The proposed protocols stack will include models for the
supported to fedora core Linux. This paper will help you channel, radio, MAC, network, transport, and higher
get started. layers.

Keywords: GloMoSim, Visualization Tool, GnuPlot and The simple approach to designing a network simulation
Parsec. would be to initialize each network node in the simulation
as a Parsec entity [6]. We can view different entity
1. INTRODUCTION initializations as being separate logical processes in the
system. Hence each entity initialization requires its own
Global Mobile Information System Simulator stack space in the runtime. In GloMoSim, we are trying to
(GloMoSim) simulates networks with up to thousand build a simulation that will scale to thousands of nodes
nodes linked by a heterogeneous communications [4]. If we have to instantiate an entity for each node in the
capability that includes multicast, asymmetric runtime, the memory requirements would increase
communications using direct satellite broadcasts, multi- dramatically [15]. The performance of the system would
hop wireless communications using ad-hoc networking, also degrade rapidly. Since there are so many entities in
and traditional Internet protocols. Developers use these the simulation, the runtime would need to constantly
simulators to model the wired or wireless network design context switch among the different entities in the system.
process [1][2]. It is being designed using the parallel This will cause significant degradation in the performance
discrete-event simulation capability provided by Parsec. of the simulation [8][9]. Hence initializing each node as a
This makes it possible to evaluate various design separate entity will inherently limit the scalability and
alterations and configurations before even deploying the performance of the simulation [16][17].
actual devices and components. Based on the outcome of
the validation process, simulations can once again be To circumvent these problems network gridding was
attempted for optimization of the hardware performance. introduced into the simulation [18]. With network
gridding, a single entity can simulate several network
nodes in the system. A separate data structure
representing the complete state of each node is
maintained within the entity. Similarly we need to
maintain the right level of abstraction. When the
simulation code of a particular node is being executed it
should not have access to the data structures of the other
nodes in the simulation. The network gridding technique
means that we can increase the number of nodes in the
system while maintaining the same number of entities in
the simulation [19]. In fact, the only requirement is that
we need only as many entities as the number of
processors on which the simulation is being run. Hence if downloaded free of cost from the URL
we are running a sequential simulation we need to [http://www.java.com/en/download/manual.jsp].
initialize only one entity in the system. We also don't
meet the memory or context switching problems that limit
2.1 Installing GloMoSim
the simulation [20]. The rest of the paper is organized as
follows. In the following section, we give the software
Before installing GloMoSim you have to install java for
required to install GloMoSim. Section 3 describes the
Linux. Simply double click jdk-1_5_0_07-linux-i586-
Visualization Tool. Section 4 describes the simulation
rpm.bin it shows at fig1 the following screen, press Run
R e s u l t a n al y sis. We draw our conclusions in
in Terminal button. It is automatically installed at the
Section 5.
directory /usr/java/jdk1.5.0_07
Fig 1: Installation of jdk1.5 in Linux
2. SOFTWARE REQUIRED TO INSTALL
GLOMOSIM After downloading the tar.gz file, simply issue the Linux
commands listed below at the command prompt, to install
To install GloMoSim Software we required following the tool.
software’s include
1.Linux of Java 1.3 or higher versions. #tar xvfz glomosim-2.03.tar.gz
2.Parsec compiler // to uncompress the glomosim software archive
3.GloMoSim software
This will create a new directory under the root directory
Usually GloMoSim software is come along with parsec named glomosim-2.03. Inside the glomosim-2.03
compiler if so no need to download parsec compiler directories, which consist of two subdirectories, can be
separately. Otherwise download parsec compiler found namely glomosim and parsec. Inside the parsec
separately. directory various Linux flavors files include freebsd-3.3,
aix, irix, redhat-6.0, redhat-7.2 and Solaris. Note that
GloMoSim tool and Parsec compiler can be downloaded downloaded software is not supported to fedora. If you
free of cost from the URL are run this tool in fedora, copy all files inside the redhat-
[http://pcl.cs.ucla.edu/projects/glomosim/] also for online 7.2 directory, paste it in /parsec directory. You can find
help refer to URL [http://pcl.cs.ucla.edu/projects/parsec/] the .makefile(make) executable file at the following
and URL [http://pcl.cs.ucla.edu/projects/glomosim]. The directory /glomosim-2.03/glomosim/main.
tool has long been available for various Linux flavors
include freebsd-3.3, aix, irix, redhat-6.0, redhat-7.2 and #cd /glomosim-2.03/glomosim/main
Solaris. Though GloMoSim are available for the
Windows platform, it is recommended you install the tool #make
on Linux. Similarly java [10] compiler for Linux can be
# vi .bash_profile
// to open the .bash_profile to customize glomosim
software
2.2 Customizing GloMoSim

As such, the path of glomosim could be anywhere on the

file system. However, it is recommended you install
glomosim on the file system of a user other than root. In
order to allow glomosim to run from the user's file directory the following: java GlomoMain. This tool
system, you may need to make a few changes to the allows to debug and verify models and scenarios; stop,
.bash_profile file under the home directory of the user. resume and stop execution; show packet transmissions,
This will also allow you to use glomosim without having show mobility groups in different colors and show
to log in as the all-powered root user. The following statistics at fig2, fig3 and fig4.
changes then should be made to the file .bash_profile. Fig 2: Visualization Tool to open bell.trace file

# User specific environment and startup programs

PATH=$PATH:$HOME/bin:/usr/java/jdk1.5.0_07/bin:/gl
omosim-2.03/glomosim/main :/glomosim-
2.03/glomosim/include:/glomosim-
2.03/glomosim/bin:/glomosim-
2.03/parsec/bin:/glomosim-2.03/parsec/include

PCC_DIRECTORY=/glomosim-2.03/parsec
export PATH PCC_DIRECTORY

It may be noted that the installation process in its final

response will suggest these changes.
Fig 3: Visualization Tool to run bell.trace file
2.3 Simulate a network using glomosim
The radio layer is displayed in the visualization tool as
After successfully installing GloMoSim, a simulation can follows: when a node transmits a packer, a yellow link is
be started by executing the following command in the drawn from this node to all nodes within its power range.
BIN subdirectory. As each node receives the packet, the link is erased and a
green line is drawn for successful reception and a red line
./glomosim inputfile > bell.trace is drawn for unsuccessful reception. No distinction is
The inputfile contains the configuration parameters for made between different packet types (ie. Control packets
the simulation (an example of such file is CONFIG.IN). A vs regular packets etc). Note that bell.trace file which is
file called GLOMO.STAT is produced at the end of the available at ./bin directory copied into ./java_gui
simulation and contains all the statistics generated. directory.

3. THE VISUALIZATION TOOL

GloMoSim has a visualization tool that is platform

independent because it is coded in java [3]. To initialize
the visualization tool, we must execute from the java_gui
 printf("Packet Received = %d \n", sumcountrcv);
printf("Packet Sent = %d \n\n", sumcountsent);
}
An example of the execution of these programs the result
as follows

#sh result1.sh glomo.stat

Loss Packet Percentage = 36.891678 %

Packet Delivery Ratio = 0.678990
Packets Received = 15233
Packets Sent = 24300

To obtain the Average Delay and the Control Packets

Fig 4: GloMoSim Statistics File to open glomo.stat file generated by the AODV routing protocol, the following
programs can be used:
4. RESULTS ANALYSIS
Shell program:
The output data of the network is generated and stored in File Name : result2.sh
glomo.stat file. In these cases we use other tools to do an
analysis of the glomo.stat file. For example, the following cat $1 | grep AppCbrServer | grep end-to-end | grep delay
shell program and awk programs is used to display packet | awk -f delay.awk
data ratio, percentage of loss rate and packets sent.
AWK program:
Shell program: File Name: delay.awk
File Name : result1.sh
BEGIN{
Cat $1 | grep App | grep Total | grep packets | awk –f sumdelay=0;
analysis.awk countdelay=0;
}
{
AWK program:
sumdelay +=$10;
File Name: analysis.awk
countdelay++;
}
BEGIN{
END{
sumcountsent = 0;
printf("Average Delay = %f \n",
sumcountrcv = 0;
sumdelay/countdelay);}
}
An example of the execution of these programs the result
{
as follows
if($10 =="sent:") sumcountsent+=$11;
else if ($10 == "received:") sumcountrcv+=$11;
#sh result2.sh glomo.stat
}
Average Delay = 12.191678 %
END{
printf("Loss Packet Percentage= %f \% \n",100-
We can even obtain some interesting graphs from simula-
((sumcountrcv*100)/sumcountsent));
tions. Lets suppose we want to obtain a graph of the vari-
printf("Packet Delivery Ratio = %f \n\n",sumcountr-
ance of packets delivered to final destination, with net-
cv/sumcountsent);
works of different size (30,50,70,100 and 200 nodes) and Perform ance
nodes move at different speeds (1,5,10 amd 20 m/s). We
can run several simulations (each with a specified node 80000
number and speed of the mobility model). To shorten the
70000
simulation time, the tests can be executed at the same
60000 30 nodes
time, thus obtaining the following table.
50000 50 nodes

Packets
40000 70 nodes
spee 30 50 70 100 200
30000 100 nodes
d nodes nodes nodes nodes nodes
20000 200 nodes
57890
10000
1 12884 34890 56899 9 67989
0
5 15927 33098 43898 37989 67700
1
10 14989 22890 34900 34900 49090
Speed (m /s)
20 12322 20323 29000 23798 29009
If the data of this table is used in a file (e.g data.txt), then
Fig 5: Packet delivery ration using Gnu Plot
we may use graphic tools such as GnuPlot [11] to obtain
an appropriate graph. In the case of GnuPlot [12], we can
GloMoSim is a popular simulation tool that is freely
execute the following commands to obtain the graph rep-
available for education, or research, or to non-profit
resented at fig5 and contained in file graph.eps.
agencies, which means you can enhance it to suit your
own requirements.
set terminal postscript eps
set size 1/1., 1/1.
set title " Performance" 5. CONCLUSION
set output "graph.eps"
set xlabel "Speed (m/s)" Detailed, high fidelity models of large networks represent
set ylabel "Packets" a significant challenge for the networking community.
plot [0:20][8000:75000] "data.txt" using 1:2 title '30 This paper presented a simulation library called
nodes' with linespoints, \ GloMoSim whose goal is to support accurate
"data.txt" using 1:3 title '50 nodes' with performance prediction of large-scale network models
linespoints, \ using parallel execution on a diverse set of parallel
"data.txt" using 1:4 title '70 nodes' with computers. The library has already been used to simulate
linespoints, \ networks with thousands of wireless nodes and provides a
"mobil.txt" using 1:5 title '100 nodes' with rich set of models for both existing and novel protocols at
linespoints, \ multiple layers of the protocol stack.
"data.txt" using 1:6 title '200 nodes' with
linespoints 6. ACKNOWLEDGEMENTS

The authors would like to express their thanks to

Dr. K. M. Mehata, the HOD and
Prof.V.M.Periasamy, the Registrar of B.S.A.
University, Chennai for the environment provided.

REFERENCE
1. GloMoSim: Global Mobile Information Systems hoc networks (dsr). Internet draft, draft-ietf-
Simulation Library. manet-dsr-09.txt,apr 2003.
http://pcl.cs.ucla.edu/projects/glomosim/ 17. Satyabrata chakrabarti and Amitabh mishra.
2. Mario Gerla Lokesh Bajaj, Mineo Takai, Rajat Ahuja, Quality of service in Mobile Ad hoc Networks.
Rajive Bagrodia. GloMoSim: A Scalable Network CRC press LLC.2003
Simulation Environment. Technical Report 990027, 18. Kui Wu and Jamelle Harms. On-Demand
University of California, 13, 1999. Multipath Routing for Mobile Ad hoc Networks.
3. Addison Lee- Kaixin Xu. GloMoSim Java Proceeding of EPMCC\ACM. 2001
Visualization Tool. Documentation version 1.1,
Software Distribution.
19. Yih-Chun Hu and David B.Johnson. Implicit
4. Rajive Bagrodia. README file – GloMoSim Source Routes for On-Demand Ad hoc Network
software. University of California, Los Angeles Routing. ACM 2001.
90095-1596. 20. IEEE 802.11: part 11: Wireless LAN Medium
5. Theodore S. Rappaport. Wireless Communications: Access control (MAC) and Physical Layer
Principles and Practice. Prentice Hall, New Jersey, (PHY) specification, Aug.1999.
1999. Author Biographies:
6. R. Bagrodia, R. Meyer et al. PARSEC: A Parallel
Simulation Environment for Complex Systems. IEEE
Computer, 98.
7. R. Bagrodia, Y. A. Chen et al. Parallel Simulation of a
High-speed Wormhole Routing Network. Proceeding
of 10th Workshop on Parallel and Distributed
Simulations, PADS 96. A.Kathirvel - born in 1976 in Erode, Tamilnadu, India, received his B.E.
8. S. Bhatt, R. Fujimoto, A. Ogieski and K. Permalla. degree from the University of Madras, Chennai, in 1998 and M.E.
Parallel Simulation Techniques for Large Scale degree from the same University in 2002. He is currently with
B.S.Abdur Raahman University, in the Department of computer science
Networks. IEEE Communication Magazine, 98, pp.
and Engineering and pursing Ph.D. degree with the Anna University,
42-47.
Chennai, India. He is a member of the ISTE. His research interests are
9. Zygmunt J. Haas et al. Wireless Ad Hoc Networks. In protocol development for wireless ad hoc networks, security in ad hoc
Encyclopedia of Telecommunications, 2002. networks.
10. Java software for Linux free download:
http://www.java.com/en/download/manual.jsp
11. Gnuplot. A Brief Manual and Tutorial. University
of Duke. www.duke.edu/~hpgavin/gnuplot.html
12. Gnuplot. Documentation version 4.0, Software
Distribution. www.gnuplot.info/ Rengaramanujam Srinivasan -- born in 1940 in Alwartirunagari,
13. Mobile Ad Hoc Networks (MANET) Working Tamilnadu, India, received B.E. degree from the University of
Madras, Chennai, India in 1962, M.E. degree from the Indian
Group,
Institute of Science, Bangalore, India in 1964 and Ph.D. degree
http://www.ietf.org/html.characters/manet-
from the Indian Institute of Technology, Kharagpur, India in
charter.html,2004 1971. He is a member of the ISTE and a Fellow of Institution of
14. Karoly Farkas, Dirk Budke, Bernhard Plattner, Engineers, India. He has over 40 years of experience in teaching
Oliver Wellnitz and Lars Wolf. QoS Extensions and research. He is presently working as a Professor of
Computer Science and Engineering at B.S.Abdur Raahman
to Mobile Ad hoc Routing Supporting Real-
University, Chennai, India and is supervising doctoral projects in
Time Applications. ACM, 2004.
the areas of data mining, wireless networks, Grid Computing,
15. C.E.Perkins, E.M. Belding-Royer, and Information Retrieval and Software Engineering.
I.D.Chakeres. Ad hoc on demand distance vector
(aodv) routing. IETF Internet draft, oct. 2003.
16. D.B. Johnson, D.A.Maltz, and Y.c.Hu. The
dynamic source routing protocol for mobile ad