A Flexible Visual Simulator for Wireless Ad-Hoc Networks of Mobile Nodes *
University of Pavia
University of Pavia
DET / IEETA - University of Aveiro
The management ofad-hoc networks raises interesting
problems, that are particularly challenging for networks
of mobile nodes.
Considering the inherent complexity
ofthese systems, the development ofdistributed applica-
tions relying on wireless communication protocols would
be greatly simplified by the use of specific tools for sup-
porting testing and step-by-step debugging.
In this paper we describe WISE, a flexible interactive
simulation environment for the development of wireless
ad-hoc networks consisting ofmobile units. A graphical
interface allows the user to create/delete nodes, change
their positions andparameters, and select specific mobil-
ity models in order to verify the network behavior in dy-
namic conditions. The simulator also provides a useful
support for the verification of agreement protocols, syn-
chronization algorithms and distributed scheduling, al-
lowing the user to display a step-by-step evolution ofthe
algorithms in a suitable graphical representation.
Keywords: wireless networks, simulation environment,
Recent advances in wireless communication technol-
ogy and a progressive reduction in the associated costs
are boosting the use of wireless networks in many di-
verse domains, either to integrate office equipment, per-
sonal equipment, or even to interconnect sensory and ac-
One of the fields in which there is strong interest in the
use of wireless networks is the interconnection of mobile
units with sensing, processing, and communication capa-
bilities for monitoring and exploration purposes.
In some cases, the possibility of controlling the po-
sition of the nodes (through teleoperated or autonomous
mobile robots) would allow much higher flexibility, be-
cause the network could be configured for a better cover-
age of the sensed area, or for actively following the evo-
lution of the phenomena [9, 12]. When nodes move, how-
*This work has been partially supported by the Italian Ministry of
University Research under contract 2003094275 (COFIN03) and con-
tract 2004095094 (COFIN04).
ever, they need to interact with the environment (e.g., to
avoid obstacles) and with the other nodes of the network;
hence, most of the activities carried out by the team need
to be executed under timing constraints, that must be en-
forced on tasks to guarantee a minimum level of perfor-
mance. In some applications the communication among
distant units is based on a wired backbone.
in the considered above situations, a wired infrastructure
cannot be used, hence a full autonomy of the team units
can only be achieved using an ad-hoc network.
Design, analysis and test ofmobile ad-hoc wireless net-
works take great advantages from specific tools. Although
some basic network and communication protocol proper-
ties can be theoretically proved, in other cases the net-
works performance and their behavior in particular condi-
tions can only be tested by simulation. In particular, this
is true for networks made by mobile units, since several
different topology configurations are possible and the net-
work connectivity changes dynamically. Moreover, the
analysis of MANETs include several issues: area cover-
age, localization, connectivity, message routing, and time-
liness in message exchanging .
Several simulation environments are currently avail-
able in the academic world. The Network Simulator ns-
2  is one of the most widely used tool for the perfor-
mance analysis and evaluation of network protocols.
recently included features to simulate wireless networks
with the introduction of the MobileNode class. However,
it cannot be used for online demonstrations and interac-
tive debugging, since it does not provide online graphical
output. The Cnet simulator  is an easy-to-use sim-
and can simulate both Ethernet based LANs and token
ring based LANs , but it does not support wireless com-
munication and does not provide online graphical out-
put. Netsim  is a powerful simulation tool, but it does
not support wireless networks, neither a graphical inter-
face. The GloMoSim  simulation environment was de-
veloped for simulating large-scale wireless networks:
is multi-platform and provides a graphical interface, but
it requires the use of the PARSEC environment  as a
parallel simulation engine to provide high computational
This paper describes WISE, Wireless Interactive Sim-
ulation Environment. WISE is a flexible visual simula-
It implements the ISO/OSI stack paradigm
0-7803-9402-X/05/$20.00© 2005 IEEE
number of nodes
Figure 1. WISE simulation levels.
tor specifically developed for ad-hoc networks of mobile
The simulator provides a basic infrastructure to
simplify both the implementation and debug ofnew algo-
rithms. WISE has been designed to be user-friendly and
can be used to analyze a protocol and to present a graphic
step-by-step evolution of the algorithm behavior. For this
purpose the package combines the simulation engine with
a highly customizable graphical interface.
The rest of the paper is organized as follows: Section
2 presents the general architecture of the simulator; Sec-
tion 3 describes the most relevant implementation details;
Section 4 illustrates the the operational modes of WISE;
Section 5 introduces the topics related with the graphic
engine; Section 6 provides an example of usage; finally,
Section 7 states our conclusions and future work.
WISE has been designed to simulate distributed pro-
tocols for wireless ad-hoc networks consisting of mobile
nodes. The main features of the simulator include:
* Native support for node mobility, to test the behavior
of the network in dynamic environments;
* Modular software structure;
* Object-oriented programming;
* Interactive graphical interface to simplify the design
of a network and the verification of network proto-
Figure 1 illustrates the hierarchial model adopted for
WISE. The model is node-centric, meaning that the Node
is the elementary component of the system. Nodes can
be grouped into one or more Networks, fully interacting
among themselves. The simulator also supports a model-
ing of the Environment where the nodes move.
Each hierarchical level is characterized by specific
properties: nodes are characterized by their position, in-
dividual behavior (i.e., moving strategy), transmission
power, etc. Networks may have a maximum number of
nodes, positioning limitations, and other constraints. The
Environment may introduce obstacles, sources of interfer-
ence and noise.
The elements of a level may affect the properties of the
elements in another level. For example, the network topol-
ogy is influenced by the nodes position; its connectivity is
affected by the presence of obstacles and by the environ-
mental interferences, which may disturb the communica-
tion between two nodes. Actually, the interaction between
a Node and the Environment is unidirectional, since the
Environment may affect the behavior of the Node through
obstacles and interferences, but the Node cannot affect the
Nodes can be freely inserted in the environment, and
they can be manually moved during the simulation in or-
der to test particular topological configurations of the net-
work. A manual control ofthe node position during a step-
by-step session can be very useful to verify the effects of
hidden terminals on the behavior of the tested algorithm.
Implementation of the model
The implementation WISE aims at achieving two levels
1. portability of the kernel on different operating sys-
2. independence among different logical modules.
The WISE core is implemented in the C++ language.
On one hand, it ensures high efficiency and strong opti-
mizations in the WISE simulation core; on the other hand,
it achieves good portability between different operating
systems. However, portability problems may arise from
the graphical interface: typically, the graphic engine is
strongly operating system dependent.
In our implementation, the simulator kernel is sepa-
rated from the graphical interface, thus allowing the simu-
lation even in the absence of graphical support. For exam-
ple, the simulation results may be collected in a text file
to be used later for off-line analysis with specific tools.
An additional advantage of separating the simulator kernel
from the graphical interface is the fact that the same sim-
ulation results can be represented under different views,
thus making their interpretation easier and immediate. For
example, two different windows may be used to separate
different representations of the same data and a third win-
dow may compare the two representations.
The kernel has been implemented by taking full advan-
tage from the object oriented hierarchical paradigm pro-
vided by the C++ language. At the bottom of the hier-
archy there is the TBasicNode class, that forms the data
background common to almost all wireless environments,
mainly concerning the node position. The TDrawNode
class extends the basic node representation by adding few
properties needed to simplify the graphic node represen-
tation. Other classes are derived from TBasicNode and
TDrawNode in order to represent specific types of nodes.
They just add the data structures and functions needed
for specialized simulations, like transmission ranges and
Every node is a member of a network. From the im-
plementation point of view, this means that the TNetwork
class is a collector of TBasicNode objects. In the same
way, TEnvironment collects a set of TNetwork classes,
since all the networks reside in the environment. New
nodes are added to existing networks, and new networks
may be added to the environment. The TNetwork and
TEnvironment containers provide the functions to deal
Figure 2. The main window of the simulator.
with their members. For example, TEnvironment provide
a function to evaluate the distance between two nodes and
this function is used to check whether a node is able to
listen to another node transmission.
Whenever a node with a new behavior is needed, the
class hierarchy has to be extended to derive the new el-
ement from the TBasicNode class. For example, a node
moving accordingly with a specific mobility model inher-
its the characteristics from a common node, but imple-
ments the new behavior for the moving strategy.
Running a simulation
The simulation goes on as a potentially infinite series
of steps, called ticks. At each tick, an event is generated
within the simulation environment, and every component
executes the corresponding action. For example, a node
may update its current position according to the mobility
model parameters, and/or it may need to broadcast a mes-
sage acting upon its communication algorithm.
WISE can run a simulation under three modes:
* in step-by-step mode;
* in live mode;
* in batch mode with (optional) online graphical out-
The three simulation modes depend on how the ticks
are generated: in step-by-step mode, each tick is gener-
ated after an explicit command issued by the user; in live
mode, the ticks are triggered periodically by an internal
timer with selectable period; finally, in batch mode, each
event associated with a tick is forthwith generated after
the end of the computation started by the previous tick.
The last mode is the fastest and may not retrieve a graphic
The three options have been proposed for different pur-
poses. The step-by-step update is useful to tune the system
parameters and to debug the algorithm. The periodic up-
date may be used to monitor the online system behavior,
Figure 3. Relationships among WISE win-
by observing the dynamics in link establishment or break-
ing together with the system data flow. The batch mode
can be used to perform simulations in which the output
data are made available for an off-line evaluation.
WISE is expressly developed to simulate networks of
mobile nodes. It actually implements the Gauss-Markov
mobility model  as well as the Random Waypoint mo-
bility model .
Moreover, the step-by-step mode al-
lows the user to manually implement an adversary-based
mobility strategy  and other models based on the cre-
ation/deletion of specific links among the nodes [8, 14].
This means that the user can opportunely change the net-
work connectivity during the simulation to easily build
specific test case situations.
The user can switch between the step-by-step mode
and the live mode at run-time. This is useful to let the
system run until a specific condition is reached and then
to analyze the system behavior more in details.
Interactive graphic interface
A good graphic feedback is one of the primary goals
of WISE. It may help to better understand the algorithm
behaviors and gives an appealing representation of proto-
col data flow and network performance. This may be used
to show complex situations in a real-time fashion, and it
is highly desirable in order to make live presentations, i.e
for teaching purposes.
The visual interface is made by a main window (shown
in Figure 2) which displays the graphic overview of the
network deployment, with options to show or hide links
among the nodes and transmission ranges (both options
are set in the example of Figure 2). By using the mouse,
each node can be freely moved in order to obtain the de-
sired network configuration, establishing new links or re-
moving existing ones. New nodes can be added in just a
mouse click. The application also supplies a set of child
windows, each one related to a specific simulation view.
Figure 3 shows the relationships among the windows
that compose the application. A number of child windows
(the simulation modules) display the evolution of the pro-
tocol data flow and every simulation module may have a
related window containing the required optional parame-
During the simulation in step-by-step and live mode,
The solution proposed in this work is WISE, a flexi- Download full-text
ble visual simulator specifically developed to interactively
display the evolution of wireless mobile networks. The
simulator natively integrates the most common used mod-
els to describe the node mobility. One of its primary goals
is a good visual feedback to simplify the comprehension
of the problem under analysis. Thanks to its user-friendly
graphical interface, WISE can be effectively used to make
live presentations, i.e. for teaching purposes.
Figure 4. Example of simulation of a dis-
tributed topology reconstruction algorithm.
the network topology represented within the main win-
dow is updated after every tick. While a step-by-step sim-
ulation is running, the nodes can be moved and the node
parameters may be changed to simulate the desired situa-
Example of usage
The WISE simulator has been profitably used to derive
many properties of the MAC level communication proto-
col presented in . WISE has been used to optimize the
algorithm and verify the convergence. Moreover, it has
been helpful to study the dynamic behavior of the method
by manually moving the nodes in order to change the net-
work topology configuration during step-by-step simula-
tion. This feature of the simulator has been particularly
useful to investigate worst-case topologies and to under-
stand the influence of link establishment/disruption on the
Figure 4 shows the output produced by the Topology
Simulation window, which displays the evolution of an al-
gorithm for the distributed network topology reconstruc-
tion. In this simulation, the nodes exchange periodic mes-
sages about their own view of the global network topol-
ogy. After every transmission, such an information is used
by the nodes which receive the data to update their own
state, making all the nodes views iteratively converge to
the correct overall network topology. Figure 4 shows the
last three steps of the topology matrix updating process,
where the matrix at i-th row and j-th column represents
the matrix owned by node i at the j-th simulation step.
The highlighted matrices identify those corresponding to
the correct global topology configuration, to simplify the
identification of the converged matrices. The same figure
also shows the window containing both the simulation and
In this paper we discussed the importance of using ap-
propriate simulation tools for the development, mainte-
nance, and testing of new distributed algorithms for wire-
less ad-hoc networks of mobile nodes.
 The NetworkSimulator
 PARSEC - The Parallel Simulation Environmentfor Com-
plex Systems. http://pcl.cs.ucla.edu/projects/parsec/.
 B. Awerbuch, P. Berenbrink, A. Brinkmann, and C. Schei-
deler. Simple routing strategies for adversarial systems.
In Proceedings of the 42nd IEEE Symposium on Founda-
tions ofComputer Science (FOCS), pages 158-167, Octo-
 L. Bajaj, M. Takai, R. Ahuja, K. Tang, R. Bagrodia, and
M. Gerla. GloMoSim: A scalable network simulation en-
vironment. Technical Report 990027, May 1999.
 B. L. Barnett.
An ethernet performance simulator for
SIGCSE Technical Symposium, pages 145-150, 1993.
 C. Bettstetter.
Smooth is better than sharp: A random
mobility model for simulation of wireless networks.
Proceedings of the 4th ACM International Workshop on
Modeling, Analysis, and Simulation of Wireless and Mo-
bile Systems (MSWiM), pages 19-27, July 2001.
 T. Facchinetti, L. Almeida, G. Buttazzo, and C. Marchini.
Real-time resource reservation protocol for wireless mo-
bile ad hoc networks. In Proceedings of the IEEE Real-
Time Systems Symposium, December 2004.
Geometric spanner for routing in mobile networks. In Pro-
ceedings of the ACM Symposium on Mobile Ad Hoc Net-
working & Computing (MobiHoc), pages 45-55, October
 Z. Haas, J. Deng, B. Liang, P. Papadimitatos, and S. Sa-
jama. Wireless ad hoc networks. December 2002.
 B. Liang and Z. J. Haas. Predictive distance-based mobil-
ity management for PCS networks. In Proceedings ofthe
18th IEEE INFOCOM, pages 21-25, March 1999.
[1 1] C. S. McDonald. A network specification language and ex-
ecution environment for undergraduate teaching. In Pro-
ceedings of the ACM Computer Science Education Tech-
nical Symposium, pages 25-34, March 1991.
 C. E. Perkins. Ad Hoc Networking. Addison Wesley Pro-
fessional, January 2001.
 D. D. Perkins and H. D. Hughes.
ity of service support in wireless ad hoc networks.
Journal of Wireless Communication & Mobile Comput-
ing (WCMC), Special Issue on Mobile Ad Hoc Network-
ing: Research, Trends, and Application(5):503-513, Au-
 M. Steenstrup.
In Proceedings of the ACM
J. Gao, L. Guibas, J. Hershberger, L. Zhang, and A. Zhu.
A survey on qual-