an extensive analysis of open source wireless network simulators · · 2014-07-08an extensive...
TRANSCRIPT
An Extensive Analysis of Open Source Wireless Network Simulators 1Koushik Banerjee, 2Hemant Sharma
1Research Scholar, ITM University, Gwalior. 2 Research Scholar, ITM University, Gwalior
E-mail: [email protected], [email protected]
Abstract
Now a day it has been observed that most of the
researchers are busy in building new simulators or
extending the existing simulation tools so that analysis of
protocols can be done and conclusions can be drawn
whether the protocols is workable in the real world. To
evaluate the performance of protocols on real network is
too much hectic in terms of cost and complexity. So
simulators are needed for reducing cost as well as
complexity. Network simulations are widely used by
researchers to test their own hypothesis and protocols on
networks without implementing in the real world. This approach saves the combination of cost, complexity and
time. There are various types of simulators available in
market among which some are general purpose simulator
others are special purpose simulators. Each simulator has
its own pros and cons therefore the researcher uses
different simulators or emulators for specific research. In
this paper, explanations and comparison of some available
network simulators like NS-2, NS-3, GloMoSim, Avrora,
COOJA, J-SIM and SENSE on the basis of multiple
parameters is presented. The aim of this paper is to identify
an effective simulator for research in Wireless sensor
network.
1. Introduction Simulation of a proposed technology is a matter of prime
importance, in order to study the impact of proposed work
a real environment is required, but testing of the proposed
technique in a real environment is too costly and restricted
in terms of flexibility, scalability etc.. Simulators provide
researchers a platform where they can design, configure,
run, analyze and draw a conclusion without implementing
their methodology on the real environment. Therefore,
researchers can easily draw a conclusion after testing their
concerned area of research. The main focus of this research
paper is to focus on parameters such as open-source,
scalability, platforms, GUI etc. For this we have designed a
comparison table which will aid researchers to easily
choose a simulator as per their requirement. The network simulators are available in a wide range which can be
compared on various parameters specifying the nodes in
term of power capacity and the links among the nodes,
traffic in the network, specifying even every tiny detail
regarding protocols, graphical user interface allows user to
visualize the movement of data packets, connectivity etc..
On the basis of simulation result one can easily analyze
their methodology.
2. RELATED WORK Researchers need a profound and analytical study for the
selection of a simulator through which they can visualize
and simulate their methodology and techniques. Different
researchers are busy testing different simulators in terms of
routing protocols [1]. Each simulator has its own merits
and demerits in terms of cost, scalability, execution
processing to evaluate the performance of network on
different parameters [2]. There are many available surveys
in the field of wireless network simulators. We have
categorized them under comparative and descriptive
papers. Some of these are provided in table-I as short
overview.
Table 1. Previous Surveys on Simulator
Ref
Number
Type Simulators
[1] comparison Opnet, NS-2, OMNeT++, SSFNet, QualNet, J-SIM, Totem
[4] description NS-2, GloMoSim, Opnet, SensorSim, J-SIM, Sense, OMNeT++, Sidh, Sens,
TOSSIM, ATEMU, Avrora, EmStar
[5] description OMNeT++, REAL, NS-2, C++Sim,
cnet, SSFNet, CLASS, SMURPH
[6] comparison SSF, SWANS, J-SIM, NCTUns, NS-
2, OMNeT++, Ptolemy, ATEMU,
Em- Star, SNAP, TOSSIM
[7] description GloMoSim, NS-2, DIANEmu, GT- NetS, J-SIM, Jane, NAB, PDNS, OM-
NeT++, Opnet, QualNet, SWANS
[8] comparison NS-2, TOSSIM
[9] comparison NS-2, Avrora, Opnet, GloMoSim
[10] comparison NS-2, cnet, JNS, Opnet, AdventNet, NCTUns
[11] comparison Opnet, NS-2
3. SIMULATORS
3.1. NS-2 NS-2[12] is a discrete event network simulator, designed
at the University of California, Berkley, which supports
wired and wireless network many network objects such as
protocols, applications and traffic source behavior. NS-2 is
open source and is a part of the software of the VINT
Koushik Banerjee et al, Int.J.Computer Technology & Applications,Vol 5 (4),1375-1379
IJCTA | July-August 2014 Available [email protected]
1375
ISSN:2229-6093
project that is supported by DARPA since 1995. NS-2 is an
object oriented simulator and more popular among
researchers. It is written in C++ and it has an OTcl
interpreter which serves as a front end. Ns-2 supports a
class hierarchy (compiled hierarchy) in C++ as well as a
very much alike class hierarchy in the OTcl interpreter.
Those hierarchies are interrelated to each other. NS-2 uses
two languages for doing two different kinds of things. NS-2
provides, detailed simulations of protocols using a systems
programming language which provide user to do
manipulation of the packet, header and bytes as well as
implementation of algorithms that run over huge data. NS-2
have reduced OSI layer model in which the presentation
and session layers are omitted. NS-2 offers a wide number
of features. Graphs can be easily generated using Xgraph or
Tracegraph. It has movement and traffic patterns and an
energy model which can be easily generated. However
traffic and mobility are typically produced before the actual
simulation start and not an integral part of NS-2’s
architecture. The main disadvantage of NS-2 is scalability
and time taking analysis. Again addition or implementation
of any new protocol is complex [11, 13]
3.2. NS-3 NS-3 is also an open source discrete-event network
simulator designed primarily for researchers to assist in
research and education. It is licensed under the GNU
GPLv2 license and freely available. The ns-3 project [14]
was started in mid of 2006 and is yet under massive
development. NS-3 is prepended as a replacement of NS-2,
not an extension [15]. NS-3 does not have an OTcl API. It
is written in C++ and python languages. In order to
simulate a methodology or technology NS-3 supports C++
and python. NS-3.20 is the latest version of NS-3. NS-3 has
enhanced characteristic and inbuilt support for parallel
simulation. NS-3 is currently supports only IPv4, which
can be considered as a limitation.
3.3. GloMoSiM GloMoSiM [16] is widely known as Global Mobile
Information System Simulator. It is a simulation
environment typically used for large scale wireless
networks. GloMoSiM is based on Parsec. GloMoSiM uses
parallel discrete-event simulation [17]. In addition, Parsec
compiler is used in GloMoSiM for compiling the
simulation of protocols [16]. The node aggregation
technique is introduced, which actually became one of its
plus points, into GloMoSim to produce significant benefits
to the simulation performance. In GloMoSim, each node
depicts a geographical area of the simulation. GloMoSim
has various choices based on its assets for CSMA MAC
protocols, wireless routing protocols and implementations
of TCP and UDP. GloMoSiM supports only wireless
network protocol.
3.4. Avrora Avrora, is an open-source, platform independent,
accurate simulator for embedded sensing programs
developed by UCLA Compilers Group. It is implemented
in Java, which helps flexibility and portability. One of the
distinguished characters of Avrora is its accuracy and
scalability. It is indeed a very powerful simulator for the
real hardware platform on which actual and real time
sensor programs run. Avrora is operating system as well as
language independent. Avrora gives an infrastructure for
analytical research as well as a framework which allows
static checking of embedded software. Avrora can
simulates a network of motes, runs the real time
microcontroller and performs actual simulations of the
devices [18]. Mica2 and MicaZ are two typical platforms
which can be easily emulated and it is also capable of
running AVR elf-assembly or binary codes. Avrora is
efficient in simulating real sensor network with accuracy. It
uses the software stack provided in TinyOS, which allow
nodes to communicate via the radio. It provides an
extension which help to simulate different sensor network
with different orientation and different number of nodes.
Developers claim that Avrora can simulate a network of up
to ten thousand nodes and its performance is very faster
than most of the other simulators [19]. On the other hand
clock drift is not modelled here, which is the major
drawback.
3.5. COOJA
COOJA [19] is a simulator which runs in the Contiki
Operating System [19] and particularly developed for
wireless networks which comprises sensors running this
particular OS. Instant Contiki [20] (a virtual machine
containing Contiki development environment) also utilize
this tool in every aspect. First things first, the simulator
operates at three layers of abstraction. The top layer is
called the networking layer, at which users not only
implement various applications but also routing protocols.
The program code which was developed and tested on
COOJA at the beginning of its formation at the operating
system level can run without any modifications directly on
Contiki OS. Layer two which is the machine code
instruction layer uses MSPSim so that the emulation at a bit
level is possible, therefore the platform is able to emulate
ESB nodes with TI MSP430 microcontrollers [21]. After
simulation, nodes can be parted in different types, both in
the software as well as hardware aspects. One more great
feature of COOJA that it also adds those sensors which do
not use Contiki. The only drawback of this tool is that one
single node is simulated entirely as one of the layers. It is
seen as a disadvantage, but the other way around, a single
network of sensors simulated at particular levels of detail
can give a broader look at the structure with an acceptable
amount of time. But there are some limitations. COOJA has
a lot of calculations and not very much efficient.
3.6. J-SIM
Java Sim or so we called J-SIM [22] is a network
simulator and built in accordance to component based
software epitome. In J-SIM culture, this is called
autonomous component architecture (ACA). J-SIM is
actually component based, and yet the compositional
simulation environment which had been developed by a
Koushik Banerjee et al, Int.J.Computer Technology & Applications,Vol 5 (4),1375-1379
IJCTA | July-August 2014 Available [email protected]
1376
ISSN:2229-6093
team at the Distributed Real Time Computing Laboratory
of the Ohio State University and Illinois University. J-SIM
is a network simulator written in Java [22]. Every dot in J-
SIM is a component. It is either a node, a link or protocol.
Work of each component is depicted using ports. Adversely
speaking, there are three possible ways to connect ports
which are one to one, one to many, and last but many to
many is more frequently used. On a more intend level, J-
SIM can be represented in two layers, The Lower layer
Core Service Layer (CSL) holds every aspect OSI layer
from network to physical where the higher layer represents
the rest of the OSI layers [23]. J-SIM is relatively complex
(not more than NS-2), inefficient and it has some overheads
in intercommunication model.
3.7. SENSE
Sensor Network Simulator and Emulator [24] or SENSE
a simulator primarily designed for sensor network
simulation. SENSE possesses a template class support
which allows the use of different components with different
types of data [25]. It can be concluded that SENSE can be
classified under component-based simulator for wireless
network simulation. SENSE is still under development
stage. When trying to use the simulator, we found some
issues that were gladly solved by the developers. This
simulator provides three user type high level, network
designers and component designers. Every component in
SENSE communicates via ports this concept free the
simulator from dependency and also enables extensibility,
reusability and scalability. Extension of components in
functionality is possible only if the interface is compatible
and no inheritance between components is used. SENSE
only supports C++ language with an interface using text,
and the results are stored in a text file. This contributes to
the efficient use of computational power, but greatly
promote adversity in the interface. SENSE is less
customizable.
4. Comparison In this comparative study the focus is given to select a
wireless network simulator as per the need so a comparison
of simulators on different criteria such as scalability, types
of network support, platform, ease of use, etc. Some
simulators are general purpose, whereas some are specific
to a particular area. Sensor researchers mainly prefer
simulators which are designed for WSN areas such as
SENSE, COOJA, J-SIM etc. However general purpose
simulators are also extending such as Mannasim patch for
NS-2 to support sensor network but scholars find learning
NS-2 too difficult. The scalability is important if one goes
for sensor network as thousands of nodes are deployed in
the network, but working on NS-2 do not provide you
scalability even GloMoSim does not support GUI for large
number of nodes but it is available free. To modify or to
test a technology simulators support particular or more than
one language so a researcher can go for a simulator for
which they have sound knowledge of languages. Table-II
provides profound analysis of simulator on a large number
of criteria. Abbreviations used in this table are listed
below:-
GUI: Graphical User Interface.
T/W: Time warp.
Ad: Ad-Hoc
5. Conclusion In this paper a comprehensive survey on different
network simulators with all background details have been
provided the study is done by keeping different parameters
in mind and their merits and demerits in particular
conditions are also discussed. Those simulators are
different, but we can say none are superior or inferior to the other. In order to select a particular simulator for a specific
research a sound knowledge of simulators is needed,
therefore the authors have given more focus on the
selection of specific simulators for particular testing to
provide clearer vision regarding Simulators in addition to a
short description, a comparative presentation is provided in
tabular form which increases more understanding. No
particular simulator is universally accepted for all
situations, therefore appropriate guidelines are mandatory
to choose simulator for a specific research.
Koushik Banerjee et al, Int.J.Computer Technology & Applications,Vol 5 (4),1375-1379
IJCTA | July-August 2014 Available [email protected]
1377
ISSN:2229-6093
Table 2. Result and Comparison among Simulators
6. References [1] S. R. Das, C.E. Perkins, E.M. Royer, “Perfor-mance
comparison of two on-demand routing protocols for ad hoc
networks,” INFOCOM, Tel-Aviv, Israel, Aug. 2000.
[2] A. R. Khan, S. A. Madani, K. Hayat, S. U. Khan,
“Clustering‐based power‐controlled routing for mobile
wireless sensor networks,” International Journal of Communication Systems, Vol. 25, No. 4, pp. 529-542, 2012.
[3] L. Begg,W. Liu, K. Pawlikowski, S. Perera, and H. Sirisena.
Survey of simulators of next generation networks for
studying service availability and resilience. Technical Report
TRCOSC 05/06, Department of Computer Science & Software Engineering, University of Canterbury,
Christchurch, New Zealand, February 2006.
[4] D. Curren. A survey of simulation in sensor networks.
Student project, www.cs.binghamton.edu/˜kang/ teaching/cs580s/david.pdf, 2007.
[5] V. Efthimia. Free tools for network simulation. Master’s
thesis, University of Macedonia, Thessaloniki, 2006.
[6] E. Egea-Lopez, J. Vales-Alonso, A. Martinez-Sala, P. Pavon-
Mari, and J. Garcia-Haro. Simulation scalability issues in
wireless sensor networks. IEEE Communications Magazine,
44(7):64–73, July 2006.
[7] L. Hogie, P. Bouvry, and F. Guinand. An overview of
manets simulation. In Electronic Notes in Theoretical
Computer Science, Proc. of 1st International Workshop on
Methods and Tools for Coordinating Concurrent, Distributed and Mobile Systems (MTCoord 2005), LNCS, pages 81–
101, Namur, Belgium, April 2005. Elsevier.
[8] M. Karl. A comparison of the architecture of network
simulators ns-2 and tossim. In Proceedings of Performance
Simulation of Algorithms and Protocols Seminar. Institut fr Parallele And Verteilte Systeme, Abteilung Verteilte
Systeme, Universit Stuttgart, 2005.
[9] A. Lemke and A. Sarkohi. Werkzeuge zur network
simulation. In G. Wittenburg, editor, Proceedings of Seminar Technische Informatik. Freie Universit Berlin, June 2006.
[10] P. Nov. Simulation of network structures. Master’s thesis,
Department of Software Engineering, Charles University in
Prague, August 2006.
[11] B. Schilling. Qualitative comparison of network simulation tools. Technical report, Institute of Parallel and Distributed
Systems (IPVS), University of Stuttgart, January 2005.
[12] The network simulator - ns-2. http:// www.isi.edu/nsnam/ns/
[13] S. Duflos, G.L. Grand, A. A. Diallo, C. Chaudet, A. Hecker,,
C. Balducelli, F. Flentge, C. Schwaegerl, and O. Seifert. “List of Available and Suitable Simulation Components”
Sl.
No
Tool Developed By License GUI Scalability Language Networ
k
Suppor
ted
Learning
curve
Emulation
Capacity
1 NS-2 University of
California,
Berkley
Open source limited
visual aid
Small C+, OTCL T/W/A
d/
WSNA
Steep No
2 NS-3 University of
Washington,
Georgia
Institute of
Technology
Open source Better Large C++ (core)
Python(bi
ndings)
T/W/A
d/WSN
A
Steep No
3 GloMoSiM Parallel
Computing
Lab, UCLA
Open source limited
visual aid
Large Parsec T/W/A
d
Moderate No
4 Avrora
UCLA
Compilers
Group
Open source Originally
command-
line
framework
Very large Java Mainly
for
WSN
Moderate Yes
5 COOJA
Adam Dunkels
Open source Good Very large Java (Simulatio
ns in C)
Mainly for
WSN
Moderate No
6 J-SIM
Ohio State
University and
Illinois
University.
Open source Good
visualizati
on
Small Java/Jacl T/W/A
d
/WSN
A
Moderate No
7 SENSE
Dept. of
Computer Science,
Rensselaer
Polytechnic
Institute
Open source Good(Usin
g G-Sense)
Large C++ Mainly
for WSN
Moderate Yes
Koushik Banerjee et al, Int.J.Computer Technology & Applications,Vol 5 (4),1375-1379
IJCTA | July-August 2014 Available [email protected]
1378
ISSN:2229-6093
Technical report, Ecole Nationale Supieure des
Tommunications (ENST), September 2006..
[14] T. R. Henderson, S. Roy, S. Floyd, and G. F. Riley, “ns-3
project goals,” workshop on ns-2: the IP network simulator,
ACM, New York, USA 2006.
[15] G. Carneiro, P. Fortuna, M. Ricardo, “FlowMonitor - a
network monitoring framework for the Network Simulator 3 (NS-3),” Fourth In-ternational ICST Conference on
Performance Evaluation Methodologies and Tools, 2009.
[16] GloMoSiM, http://pcl.cs.ucla.edu/projects/glomosim/, ac-cessed Sept 20th 2011.
[17] R. Bagrodia, R. Meyerr, “PARSEC: A Parallel Simulation
Environment for Complex System” UCLA technical report,
1997.
[18] Ben L. Titzer, Daniel K. Lee, Jens Palsberg, “Avrora: Scalable sensor network simulation with precise timing,” 4th
Int. Conf. on Information Processing in Sensor Networks,
2005.
[19] F. Osterlind, A. Dunkels, J. Eriksson, N. Finne, and T. Voigt, “Cross-level sensor network simulation with COOJA,” 1st
IEEE International Workshop on Practical Issues in Building
Sensor Network Applications, pp. 8, Tampa, Florida, USA,
2006.
[20] “Instant Contiki.” http://www.contiki-os.org/p/instant-contiki.html accessed June 2012.
[21] T. Voigt, J. Eriksson, F. ¨ Osterlind, R. Sauter, N.
Aschenbruck, P. J. Marr´on, V. Reynolds, O. V. L. Shu, A.
Koubaa, and A. K¨opke, “Towards Comparable Simulations of Cooperating Objects and Wireless Sensor Networks,” in
Proceedings of WSNperf, 1st International Workshop on
Performance Methodologies and Tools for Wireless Sensor
Networks, (Pisa, Italy), 23 October 2009.
[22] J-Sim user, Google Groups. “http:// Groups. Google.com/group/J-Sim-users/browse_threds”
[23] M. M. Koksal “A survey of network simulator supporting
wireless networks.” Middle East technical university Ankara,
Turkey, pp.1-11, 2008.
[24] Chen, G., J. Branch, M. J. Pflug, L. Zhu and B. Szymanski
(2004). SENSE: A Sensor Network Simulator. Advances in
Pervasive Computing and Networking. B. Szymanski and B.
Yener, Springer: 249-267.
[25] B.K. Szymanski and G.G. Chen. Sensor Network Component
Based
Simulator. Handbook of Dynamic System
Authors Biography
Koushik Banerjee received his B.Tech degree from St. Thomas
College of Engineering & Technology,
Kolkata in 2011. He is currently pursuing M.Tech from ITM University, Gwalior. He is
very much attracted towards computer
programming. His research interest includes protocols and secure routing in
wireless sensor network as well as other Ad-Hoc networks. He is also interested in
different aspects of cloud computing.
Hemant Sharma received his B.E degree
from Maharana Pratap College of
Technology, Gwalior in 2011. He is
currently pursuing M.Tech from ITM University, Gwalior. He is interested in
security issues in sensor network, cloud
computing.
Koushik Banerjee et al, Int.J.Computer Technology & Applications,Vol 5 (4),1375-1379
IJCTA | July-August 2014 Available [email protected]
1379
ISSN:2229-6093