Democratic and Popular Algerian Republic

Ministry of Higher Education and Scientific Research

Master's Thesis: Computer Science

option: Analysis, Command and Monitoring of Systems
Theme:

Graphic Analysis for monitoring in a

wireless sensor network (WSN)
Simulator: OMNET++

By: Leila Imane NIAR

Supported in ..JULY 2012 ..before the jury composed of

Mr. Mustapha Kamel RAHMOUNI President

Mr. Bouabdellah KECHAR Examiner
Mr. Mejdi KADDOUR Examiner
Mr. Mohammed FEHAM Examiner
Mr. Hafid HAFFAF Counselor
 Thank you

My thanks go to my supervisor H. HAFFAF, for

his help, his encouragements, and his constructive criticisms that have
greatly helped to appreciate this work and to better illuminate my perspectives
I am grateful to him, particularly for the trust.
that he made me.

I would like to thank the jury members for giving me the pleasure
to agree to examine this work.

I would also like to thank my family and my friends.

vail would certainly never have seen the light of day without their help, and their
I would like to sincerely thank my supporters.
 ii

Dedication

For Father, Mother, Sister, and brothers

TableofContents

Summary: thirteen

General Introduction xiv

1 Presentation of Wireless Sensor Networks 2

1.1 Wireless Networks 4
1.1.1 The wireless personal network ...
4
1.1.2 The wireless local area network (WLAN) 5
1.1.3 The metropolitan wireless network (WMAN) .6. .
1.1.4 The wireless wide area network (WWAN) 6
1.2 Ad Hoc networks 7
1.2.1 Characteristics of Ad Hoc Networks 7
1.3 Wireless Sensor Networks (WSN) ...
7
1.3.1 Architecture of a sensor node . . .8 . . . . .
1.3.2 Organization of a sensor node... 9
1.3.3 Architecture of a sensor network . . . . . . 12
1.3.4 The types of RCSF architectures . . . . . . .
1.3.5 The different topologies of RCSF: . . . . . . 1
[Link] The Star Topology: . . . . . . . .
[Link] Grid topology : . . . . . . . . .
[Link] The hybrid topology: . . . . . . . .
1.3.6 The information collection: . . . . . . . . . .
[Link] On request: . . . . . . . . . . .
[Link] Following an event: . . . . . . . .
TABLE OFCONTENTS iv

1.3.7 The main characteristics of SRFC . . . 17

1.3.8 Areas of application . . . . . . . . . . . .
1.3.9 Energy consumption in RCSF . . . 19
[Link] The main operations due to the
energy consumption . . . . . . . . 19
[Link] Energy consumption model . . 21
[Link] The factors involved in consumption
energy mation . . . . . . . . . . . . . 22
1.3.10 Routing in the RCSF . . . . . . . . . . . . . 24
[Link] Examples of routing protocols . . 24
1.4 Conclusion . . . . . . . . . . . . . . . . . . .

2 The monitoring of Wireless Sensor Networks 31

2.1 The supervision of the RCSF . . . . . . . . . . . . .
2.1.1 The Breakdowns . . . . . . . . . . . . . . .
[Link] Classification of failures . . . . . . .
2.1.2 Fault-tolerant architectural solution in
the RCSF . . . . . . . . . . . . . . . . . . . . . . . 35
2.1.3 Monitoring aspect of an RCSF . . . . . . . .
[Link] Méthodologie de déploiement : . . . . 37
[Link] Network Connectivity . . . . . . . . .
[Link] The coverage . . . . . . . . . . .
[Link] Longevity of the network . . . . . . .
2.1.4 Redundancy . . . . . . . . . . . . . . . .
[Link] Redundancy Material . . . . . . .
[Link] Analytical redundancy . . . . . . .
2.2 Multi-agent Systems . . . . . . . . . . . . . . . . .
2.2.1 Characteristics of Multi-Agent Systems . . 43
2.3 Les Travaux Antérieurs . . . . . . . . . . . . . . . .
2.4 Description of the distributed (decentralized) approach . . 47
2.4.1 Network topology . . . . . . . . . . . . . .
TABLE OFCONTENTS v

2.4.2 Definition of roles . . . . . . . . . . . . .

[Link] Role Assignment Algorithm . . . . 49
2.5 The adopted routing protocol . . . . . . . . . . . . .
2.5.1 The UML Diagrams . . . . . . . . . . . .
[Link] Neighborhood Detection . . . . . . . 52
2.5.2 Role change: . . . . . . . . . . . . . . .
2.5.3 Detection of neighboring groups . . . . . . . .
2.5.4 Verification of group consistency: . . . . 54
2.5.5 Resolution of conflicts between representatives . . . 55
2.5.6 Management of Redundancy . . . . . . . . . .
[Link] The principle followed to manage the redirection
dance : . . . . . . . . . . . . . . . . . . . 57
[Link] Fault detection mechanism. 58
2.5.7 Information collection . . . . . . . . . . . .
2.6 Conclusion . . . . . . . . . . . . . . . . . . . . .

3 Simulation of monitoring in RCSF 62

3.1 General information on simulation . . . . . . . . . . . .
3.2 Types of simulation . . . . . . . . . . . . . . . . . . .
3.2.1 Discrete simulation systems: . . . . . . . . . 64
3.2.2 Continuous simulation systems : . . . . . . . .
3.3 The existing network simulators . . . . . . . . . . . .
3.3.1 NS2 : . . . . . . . . . . . . . . . . .
3.3.2 GloMoSim . . . . . . . . . . . . . . . .
3.3.3 OMNET++ ................................... 68
3.4 Comparison between the simulators . . . . . . . . . . .
3.5 The OMNET++ Simulator (Objective Modular Network
Testbed in C++) . . . . . . . . . . . . . . . . . . . . . . . 70
3.5.1 Choice of the OMNET++ simulator . . . . . . . .
3.5.2 Presentation of OMNET++ . . . . . . . . . .
3.5.3 Architectural description of OMNET++ . . . . . 71
TABLE DES MATIÈRES vi

3.5.4 Installation of the OMNET++ simulator . . . . . .

3.5.5 The main files of OMNET++ . . . . . . 73
[Link] File (.Ned): . . . . . . . . . . .
[Link] File (.ini): . . . . . . . . . . . .
[Link] File (.msg): . . . . . . . . . . .
3.6 The OMNET++ platforms . . . . . . . . . . . . . . 7
3.6.1 Mobility Framework . . . . . . . . . . . . .
[Link] The structure of a mobile host . . . .
3.6.2 Mixim .................. 78
3.6.3 Castalia
3.7 Details on Castalia . . . . . . . . . . . . . . . . .
3.7.1 The MAC module . . . . . . . . . . . . .
3.7.2 The Radio module . . . . . . . . . . . . .
3.7.3 Wireless channel . . . . . . . . . . . . . .
3.7.4 The Routing module . . . . . . . . . . . .
3.7.5 The Application module . . . . . . . . . . .
3.8 Outils de Simulation . . . . . . . . . . . . . . . .
3.8.1 Installation of OMNET++ Version 4 .......... 83
3.8.2 Installation of Castalia 3.1 . . . . . . . . . .
[Link] The Castalia Commands . . . . . . .
3.9 Implementation of a network . . . . . . . . . . . . .
3.10 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 86

4 Implementation and Simulation 88

4.1 Environnement de travail . . . . . . . . . . . . . . .
4.1.1 Hardware Environment . . . . . . . . . . . .
4.1.2 Software Environment . . . . . . . . . . . .
4.2 Simulation process . . . . . . . . . . . . . . . . . .
4.3 Project Implementation . . . . . . . . . . . . . . . .
4.3.1 Objective . . . . . . . . . . . . . . . . .
4.3.2 Description of the design of a node . . . . 91
TABLE OFCONTENTS vii

4.3.3 Description of the network . . . . . . . . . . .

4.4 Implementation . . . . . . . . . . . . . . . . . . .
4.4.1 Structure of the messages used . . . . . . . .
[Link] At the Application layer . . 96
[Link] At the Network layer . . . . 96
4.4.2 Overview of Source Codes . . . . . . . . . .
[Link] Periodic Sending of the HELLO Message
98
[Link] Attribution Des Rôles . . . . . . . .
[Link] Redundancy Management . . . . . . . .
[Link] Rebroadcast of Hello . . . . . . . .
4.4.3 The .ini file . . . . . . . . . . . . . . .
4.5 Résultats Obtenus . . . . . . . . . . . . . . . . . .
4.5.1 Network Deployment . . . . . . . . . . . . .
[Link] Node Initialization . . . . . . . . 105
[Link] Neighborhood Recognition . . . . . . 106
[Link] Neighborhood Table and Assignment
roles . . . . . . . . . . . . . . . . . . . . 108
[Link] Transition to Sleep State . . . . . . .
4.5.2 Example of a Resume file . . . . . . . . .
4.5.3 Simulation Parameters . . . . . . . . . . . . 1
[Link] The Radio CC1000 . . . . . . . . .
4.6 Results . . . . . . . . . . . . . . . . . . . . . .
4.6.1 Network Organization . . . . . . . . . . . . .
[Link] The positions of the nodes . . . . . .
[Link] The neighborhood . . . . . . . . . .
4.6.2 Node Redundancy . . . . . . . . . . . . . .
4.6.3 Network Reconfiguration . . . . . . . . . . . .
4.6.4 Energy Consumed at the Network Level . . . . 11
4.6.5 Energy Consumed With and Without Monitoring. 117
4.6.6 Remaining energy of a node Representing . . . .
TABLE DES MATIÈRES eight

4.6.7 Average number of packets from different nodes

4.6.8 Energy Consumed by the Different Nodes . . . . 1
4.6.9 Influence of Node Mobility on the Network
4.6.10 Influence of Simulation Time . . . . . . . . 122
4.7 Conclusion

General Conclusion 124

TABLE OFCONTENTS nine
Tableoffigures

1.1 The categories of wireless networks. [MAT08] . . 4. . . .

1.2 the sensor node [CAS08]. 8
1.3 the components of a sensor node [MOA08]. . .8 . . .
1.4 The Protocole Stack [CSS04]. . . . . . . . . . . . .
1.5 Communication architecture of an RCSF [KEC07]. . . 12
1.6 Architecture Plat of the RCSF[ROM07]. . . . . . . . . . .
1.7 Hierarchical architecture of RCSF[ROM07]. . . . . . 14
1.8 Collection on demand . . . . . . . . . . . . . . . .
1.9 Collection following an event . . . . . . . . . . . . .
1.10 The various applications of RCSF [CHE08]. . . . . 20
1.11 Energy consumption model . . . . . . . . . . . 21
1.12 The listening in [MOA08]..................... 23
1.13 Classification of routing protocols [BEN09] . . . 25
1.14 The steps of the SPIN protocol[CHE08]...
1.15 Steps describing the DirectedDiffusion protocol [CHE08].
28

2.1 Classification of failures [CHE08]. . . . . . . . . . . .

2.2 Strategies for deploying nodes in a RCSF[YAK].
38
2.3 Hardware Redundancy [RIP99] . . . . . . . . . . . . 4
2.4 Organization in regions of sensor nodes [ALL09]. 48
2.5 Allocation of roles in an RCSF. . . . . . . . . . . .
2.6 State Transition Diagram........................................51
TABLE OFFIGURES xi

2.7 Sequence Diagrams.................................... 52

2.8 Introduction of a new node. . . . . . . . . . . . . 53
2.9 Detection of adjacent groups. . . . . . . . . . . . . .
2.10 Correction of the inconsistency problem. . . . . . . . . . 55
2.11 Conflict resolution between Representatives...
2.12 Redundancy of simple nodes. . . . . . . . . . . . . . 57
2.13 State change of a node............................. 58

3.1 Architectural description of the NS2 simulator. . . . . . 66

3.2 Packet transfer in GloMoSim.................. 67
3.3 The launch of the Omnet++ simulator . . . . . . . . .
3.4 Modular architecture of the Omnet++ simulator . . . . 72
3.5 NED file in graphic mode. . . . . . . . . . . . . .
3.6 NED file in text mode. . . . . . . . . . . . . . . .
3.7 Example of a *.Ini File. . . . . . . . . . . . . . .
3.8 Executing a simulation in OMNeT++.
3.9 Mobility and Channel Control Architecture. . . . . . . 77
3.10 Internal architecture of a node in MF. . . . . . . . . 78
3.11 The connections of the modules under Castalia. . . . . . . . 79
3.12 The main components of a node . . . . . . . . . . 80
3.13 The finite state machine of the Radio module. . . . . . . . . . 81

4.1 Simulation Process..................................... 91

4.2 Graphic file '.ned' of the node. . . . . . . . . . . . .
4.3 Graphic file '.ned' of the network. . . . . . . . . . . .
4.4 '.msg' file of the message Data................................... 96
4.5 The structure of the Hello and HelloRep message. . . . . . .
4.6 The structure of the message 'Pass Representative' and 'Re-
conflict resolution
4.7 The structure of the messageVerification of consistency and
Change of role................. 98
4.8 Redundancy chart of simple nodes... 101
TABLE OFFIGURES twelve

4.9 Deployment of nodes.......................... 104

4.10 Execution with the Deployment . . . . . . . . . . . . . . 105
4.11 Random Deployment. . . . . . . . . . . . . . . . . . . 113
4.12 Organization of Nodes.......................... 114
4.13 Redundancy Management. . . . . . . . . . . . . . . . . 115
4.14 Reconfiguration following a failure......... 115
4.15 Energy Consumed in Network. . . . . . . . . . . . . . . 116
4.16 Energy Consumed at the Level of the Two Methods . . 117
4.17 Average Energy Consumed at 50 knots . . . . . . 117
4.18 Average Energy Consumed per 100 knots . . . . . 118
4.19 The remaining energy at a node Representing . 118
4.20 Packets sent and received by nodes. . . . . . . 119
4.21 Energy Consumed With Monitoring. . . . . . . . . . 120
4.22 Energy Consumed Without Supervision. . . . . . . . . . 120
4.23 Influence of Mobility on the Network. . . . . . . . . . . 121
4.24 Influence of Mobility on the Network (Average Energy).121
4.25 Remaining Energy in relation to Simulation Time. . 122
4.26 Average Remaining Energy. . . . . . . . . . . . . . . . . 122
Summary:

Wireless sensor networks attract more attention from the com-

research community. Thus, simulation is an approach
commonly used to test and validate approaches, the environments
simulation environments must be able to support the various mo-
wireless sensor networks. Our article focuses more
specifically in the case of monitoring sensor networks. The proto-
The tested school is a protocol based on the reconfiguration of nodes with
the principle of redundancy to maintain coverage of the area and
ensuring the long life of the network. The validation of the approach is
considered following the use of the OMNeT++ simulator since the
the results obtained showed that the concept of the approach at the level
from fault tolerance (with the principle of waking up nodes
asleep in case of failure or energy exhaustion of a node
tif) ensures good operation as well as longevity of the network.

Mots Clés : Réseau de capteurs sans fil, Surveillance, Redon-

dance, Reconfiguration, Simulation, OMNET++, Castalia
General Introduction

The advancement of technologies in network infrastructures

and the small sensors of the network allows many ap-
sensor network applications ranging from civilian to military,
house in the environment and nature from the industrial to the domain
commercial growth is increasing, as nowadays the need to observe
physical phenomena such as temperature, pressure, or even
Brightness has become essential. Examples include monitoring of
the habitat of animals, observation of the environment and prevention
sion, the human body the tracking, the battlefield of detection and
of analysis, etc

However, the reduced size of the sensors requires the use of a

battery as a source of energy which implies a limited duration of
network. Following that, the collaboration of a number of these nodes provides
birth to a wireless sensor network. The latter is deployed from
randomly in a field.

The sensors are designed for information collection according to

very specific criteria. Thus, energy consumption has become a
major axis in research from which several routing protocols originate
were proposed with the aim of optimizing energy consumption
Techniques and extending the life span of the network.
General Introduction fifteen

A routing protocol allows the routing of information

within the network with the sending of messages between sensor nodes
for the collection of the phenomenon. This turns out to be energy-intensive.
causes battery depletion, even more data loss
and the reduction of network life.

Dans ce contexte, la fonction de surveillance en continue est une

method based on the detection and localization of anomalies which
can occur in the network.
Following this, one of the promising solutions is the use of
design of a multi-Agent system with the principle of monitoring.
The objective of our work is to create in the simulation environment
OMNeT++, a network monitoring model. This model
is based on distributed management with the concept of redundancy.
As a consequence of this surveillance: The decrease of my-
["common messages","The network hierarchy","Energy reduction"]
consumed and thus the longevity of the network.

Following this, our document is organized into 4 chapters:

-Chapter 1: Provides an overview of wireless sensor networks
Files and their characteristics. A state of the art is presented.

Chapter 2: Is dedicated to network monitoring where we

presented the main concepts of monitoring and more
precisely at the level of wireless sensor networks. The
works in this field are cited as references.

Chapter 3: Presents the contribution of simulation for validation

Comparison of results. A comparison between the different simulators.
will enrich the content of the chapter. Our choice will be justified.
is focused on OMNeT++ and Castalia.
General Introduction sixteen

Chapter 4: Constitutes the heart of this contribution in proposing

without a network monitoring model as well as its implementation
simulation. The simulation results will also be presented
tests to justify the validity of our approach.
General Introduction 1
Chapter 1

Presentation of Sensor Networks

wireless

During its evolution, the wireless paradigm has seen the birth of di-
derived architectures, such as:
cellular networks,
wireless local area networks,
– etc..
During the last decade, a new architecture has emerged:
wireless sensor networks (WSN).

These have been used in numerous applications such

that the monitoring of forests, disaster management, the exploration-
spatial quality, industrial automation, installation of locks,
border protection and battlefield surveillance
[ASS08][CKU03].

In these applications, the miniaturized sensor nodes are...

employees functioning autonomously in environments
without supervision. In addition to the ability to explore its environment.
Each sensor has an onboard radio used for transmission.
data collected at a base station, either directly or at
Presentation of Wireless Sensor Networks 3

through a multi-hop path. These sensors have 3 functions:

1. Capture data (such as sound, vibration, light,...)
2. Calculate information using these collected values,
3. Communicate them through a sensor network.

This random dispersion of sensors requires a protocol based

on self-organization algorithms. In order to resist deployments-
To be effective, these sensors must be very robust and furthermore, they must
also be able to survive in the most extreme conditions dictated
by their environment of use (fire or water for example). In
more environmental constraints, a very important constraint
it is the battery economy. Indeed, a network of sensors does not
can survive if the loss of nodes is too significant because this in-
generate communication losses due to a great distance
between the sensors. So it is very important that the batteries last
as long as possible since in most of the ap-
In applications, the sensors are placed randomly.
Presentation of Wireless Sensor Networks 4

1.1 Wireless Networks

There are several categories of wireless networks that differ by the

geographical perimeter they cover as well as by the types of ap-
supported applications. The following diagram illustrates the categories of
wireless networks.

FIG. 1.1 – The categories of wireless networks. [MAT08]

1.1.1 The wireless personal network

It concerns low-range wireless networks: on the order of

a few dozen meters. This type of network is generally used to
connect devices (printer, mobile phone, devices
domestics, PDA...). There are several technologies used for
the WPAN such as:

1. Bluetooth technology: It is also known as

IEEE 802.15.1 standard, it was launched by Ericsson in 1994, pro-
putting a theoretical flow rate of 1 Mbps allowing it to trans-
mission of voice, data, and images [2], of a range
Presentation of Wireless Sensor Networks 5

maximum of about thirty meters.

Bluetooth is a low-cost technology, thanks to its strong in-
integration on a single chip of 9 mm by 9 mm [3]; It pre-
also has the advantage of working on devices with
low power resulting in low energy consumption [1].
2. ZigBee technology: It is also known as the standard
IEEE 802.15.4 allows for low-cost wireless connections.
with a very low energy consumption, which makes it
particularly suited to be directly integrated into
small electronic devices (sensors, household appliances
gers...) [1].
ZigBee networks allow for data rates of up to 250
Kbits/s in the classic 2.4GHz band. The RCSF constit-
one of the applications that this standard can cover [1].
3. Infrared connections: Allow for the creation of wireless connections.
a few meters with flow rates that can reach several
megabits per second. This technology is widely used
in home automation (remote controls), and nevertheless suffers from per-
perturbations caused by light interference.

1.1.2 The wireless local area network (WLAN)

It is a network that allows covering a range of about a

hundred meters. It connects the terminals to each other pre-
sent in the coverage area. There are two competing technologies.
rents:
1. Wi-Fi networks (Wireless-Fidelity): Come from the standard
IEEE 802.11, which defines a cellular architecture. It includes
mainly two types of wireless networks: Those that tra-
operate at a speed of 11 Mbits/s at 2.4 GHz (IEEE 802.11b) and
those who reach 54 Mbits/s at 5 GHz (IEEE 802.11 a/g).
Presentation of Wireless Sensor Networks 6

2. The HiperLAN 2 networks (High Performance LAN 2.0): Result from

from the European standard developed by ETSI (European Telecom-
HiperLAN 2 allows for
a theoretical speed of 54 Mbps over an area of about a hundred
meters in the frequency range between 5 150 and 5 300
MHz [1]. This type of network has not received as much success as the
Wi-Fi technology.

1.1.3 The metropolitan wireless network (WMAN)

Also known as Local Radio Loop (LRL). It is suitable

to recall that the BLR allows, by placing a satellite dish
on the roof of a building, to transmit via radio waves
voice and high-speed data for internet and telephony access
No. There are several types of WMAN networks, the most well-known of which is
east
The Wimax network (Worldwide interoperability for Microwave Ac-
cess) :
They stem from the IEEE 802.16 standard and aim to develop
for competing Hertzian links to xDSL techniques
rest and offer a useful bandwidth of 1 to 10 Mbit/s in the 10-66 band
GHz for a range of 4 to 10 kilometers, which primarily targets
teaches this technology to telecommunications operators.

1.1.4 The wireless wide area network (WWAN)

It is known as the mobile cellular network and it is the most

widespread since all mobile phones are connected to a re-
extended wireless bucket. The main technologies are as follows:
GSM (Global System for Mobile Communication), GPRS (General Packet Radio Service)

Packet Radio Service), UMTS (Universal Mobile Telecommunication

System).
Presentation of Wireless Sensor Networks 7

1.2 Ad Hoc networks

An ad hoc network, or MANET (Mobile Ad hoc NETwork), is a

dynamically formed network by an arbitrary set of nodes
independent. No pre-installation regarding the role that should
playing each node is required.
In an ad hoc network, the nodes are assumed to behave like
routers and clients at the same time, also, each node is free to
move and randomly organize themselves. Thus, the topology of the re-
The sea can change quickly and unpredictably.

1.2.1 Characteristics of Ad Hoc networks

Node Mobility: In an ad hoc network, the topology of the re-

The basket is dynamic and can therefore change quite rapidly.
Wireless connections: The only means of communication in the...
Ad hoc buckets are the use of wireless interfaces. These connections
wireless will always have performance inferior to their wired counterparts
wired cables [CAM99].
Node equivalence: In Ad Hoc networks, there is none.
of difference between nodes such as in other networks (host and sta-
for all nodes can be made to perform functions
routing.

1.3 Wireless Sensor Networks (WSN)

Sensor networks are considered a special type

of Ad hoc networks. As a result, they inherit characteristics
of these networks, including the architecture without established infrastructure and
wireless communication. The RCSF form a new generation.
network design with specific properties, They present a field
Presentation of Wireless Sensor Networks 8

of very broad application and cover several areas of a character

scientific, logistical, military or health.

An RCSF is made up of several hundred and sometimes thousands

sensor nodes, each node is capable of monitoring its
environment and to react if needed by sending the information
information collected at one or more collection points, using a
wireless connection [BAB06].

FIG. 1.2 – the sensor node [CAS08].

1.3.1 Architecture of a sensor node

A sensor node consists of four main units, which

are presented in the figure below [CSS04][KHA06].

FIG. 1.3 - the components of a sensor node [MOA08].

a. Sensing unit: It is composed of two sub

units, a physical capture device that collects information-
Presentation of Wireless Sensor Networks 9

local environment management and an analog converter/

digital called ADC (Analog to Digital Converters).
b. Processing unit: The captured data is
communicated to the processor where they are stored in the me-
moire.
c. Communication unit (Transceiver unit): It is composed of
of a transmitter/receiver (radio module) enabling communication
communication between the different nodes of the network.

d. Energy unit (Power unit): It is the battery that is generally...

non-rechargeable and non-replaceable element. Energy capacity li-
mitigated at the level of the sensors represents the main constraint
during the design of protocols for sensor networks.
Energy units can be supported by solar cells.
which allow converting light energy into electric current
trick.

1.3.2 Organization of a sensor node

sensor networks use a communication protocol stack

communication composed of five layers: an application layer, a
transport layer, a network layer, a data link layer
and a physical layer. This stack is also characterized by three
levels that integrate into the protocols of the different layers the
consideration of the energy consumed, the mobility of the nodes,
and the management of task distribution across different nodes
of the network [DOH06].

a. The physical layer: It is responsible for frequency selection.

sequence, the generation of the carrier frequency, the detection of
signal, modulation/demodulation and encryption/decryption
information. It is advantageous in terms of energy economy.
Presentation of Wireless Sensor Networks 10

FIG. 1.4 – The protocol stack [CSS04].

when the physical layer designer chooses a trans-

multi-hop mission rather than a direct transmission that ne-
requires a very high transmission power [CSS04].
b. The data link layer: The data link layer is
mainly responsible for:
Multiplex the data stream.
Detect and lock data frames.
Control access to the transmission medium (Media Access
Control).
Control the errors.
– And to ensure a reliable connection (point-to-point or point-to-
multipoints) according to the topology of the sensor network.

c. The network layer: Manages exchanges (and possibly connections)

through the RCSF. Knowing that the positioning of the nodes
being random and dense, traditional routing protocols
become inadequate. For this, multi-hop communication
is the best suited. This layer supports the cutting-
Presentation of Wireless Sensor Networks 11

neighborhood view, resource allocation, and routing.

d. The transport layer: In sensor networks, the layer
transport is essentially present to constitute an inter-
face between the application layer and the network layer [HAM07].
Its main objectives are:
Multiplex and demultiplex messages between the applications
tions and the network layer.
Control high-level data.
Regulate the amount of data injected into the network.
The role of this layer mainly comes into play when we go
access from our RCSF to another RCSF or to Inter-
net.
The application layer: The application layer constitutes the...
set of applications implemented on a sensor network.
These applications should provide mechanisms allowing for
the user to interact with the sensor network through dif-
reference interfaces, and possibly, through a
wide area network (for example: Internet).
This layer is responsible for example for the collection, the co-
Therefore, the aggregation and compression of the collected data.
Presentation of Wireless Sensor Networks 12

In addition to the 3 levels (plans):

a. The energy management level: Responsible for controlling the way

a node does not use its energy.
b. The level of task management: Ensures the balancing of the dis-
task distribution on the different nodes to accomplish a
cooperative work.
c. The level of mobility management: Detects and records everything
the movements of the sensor nodes.

1.3.3 Architecture of a sensor network

Sensor nodes are usually scattered in an area.

of capture. Each of these nodes has the ability to collect the data
born and route them to one or more base stations (sink
node). The latter is a data collection point for captured data. It
can communicate the collected data to the end user through
towards a communication network, possibly the Internet[MOA08].

FIG. 1.5 – Communication architecture of a RCSF [KEC07].

As shown in (Figure 1.5), a RCSF is composed of a large

number of sensor nodes scattered across the catchment area [ELK].
Presentation of Wireless Sensor Networks 13

At a higher level, a RCSF can be seen as a

combination of two network entities:
The data acquisition network: It is the union of the nodes.
sensors and the sink, its role is to collect data from
start from the environment and gather them at the sink.
The data distribution network: Its role is to connect
to link the data acquisition network to a user.

1.3.4 Types of RCSF architectures

1. Flat wireless sensor networks: A network of sensors without

fil plat is a homogeneous network, where all nodes have
of the same capabilities in communication, capturing information-
and a different sink since it plays the role of a gateway
responsible for transmitting the collected information to the user.

FIG. 1.6 – Architecture Plan of the RCSF[ROM07].

2. Hierarchical wireless sensor networks: It is a network

heterogeneous where the nodes can have a power source
critique, of a range of communication or of a power of
calculates differently from each other.
Presentation of Wireless Sensor Networks 14

FIG. 1.7 - Hierarchical architecture of RCSF[ROM07].

1.3.5 The different topologies of RFCs:

[Link] The Star Topology:

In this topology, a base station can send or receive

a message to a certain number of nodes. These nodes can
only send or receive a message from the unique station of
Basically, they are not allowed to exchange messages.
The advantage of this topology is its simplicity, its ability to minimize
the energy consumption of the nodes and the minimization of the
sentence of communication between the nodes and the base station. Its
the disadvantage is that the base station is not robust since everything
the network is managed by a single node.

[Link] The grid topology:

In this type of topology, any node can send to

any other node in the network that is in range of
transmission. This is called multi-hop communication, in the-
which, if a node wants to transmit a message to another node that
is outside its transmission range, it uses an inter-node
medium to send its message to the destination node. The advance-
Presentation of Wireless Sensor Networks 15

The advantage of this topology is the possibility of scaling up, the re-
dancing and tolerance to mistakes, The drawback of this topology
is the energy consumption in multi-hop communication and
the latency created by the passage of messages between nodes
before arriving at the base station.

[Link] The hybrid topology :

A hybrid topology between star and grid provides

robust and diverse network communications, ensuring the minimum
reduction of energy consumption in sensor networks.
In this type of topology, low-power sensor nodes do not
do not route the messages, but there are other nodes that have the poss-
ability to route messages. In general, these nodes have
a high power.

1.3.6 The collection of information:

There are two methods to collect information from a network

of sensors:

[Link] On request:

When one wants to have the status of the coverage area at a

At that moment, the well broadcasts to the entire area so that the
sensors send their last readings to the well. The information
are then routed through a multi-hop communication.

[Link] Following an event:

An event occurs at a point in the coverage area.

(sudden change in temperature, movement...), the sensors if-
Presentation of Wireless Sensor Networks 16

FIG. 1.8 – On-demand collection.

killed nearby then report the information collected and the buy-
leading up to the well.

FIG. 1.9 - Collection following an event.

Presentation of Wireless Sensor Networks 17

1.3.7 The main characteristics of RCSF

Reduced energy consumption: Each node has a

small-sized battery, and is therefore exposed to "death"
once all its energy stock is depleted.

Node sensor self-configuration: In a RCSF, the

nodes are deployed in a random manner (missile, plane...)
Thus, a sensor node must have capabilities of a
part, to self-configure in the network, and on the other hand for
collaborate with other nodes in order to reconfigure dy-
dynamically the network in case of topology change of the
network [HOW03].

Scalability: A sensor network is scalable because it has the

ability to accept a very large number of nodes.

Tolerance to failures: In the event of a malfunction

of a node (lack of energy, interference with the environment-
observation period...) or also in case of adding new nodes
sensors in the network, this node must continue to function
normally without interruption. This explains the fact that a RCSF
does not adopt a fixed topology but rather a dynamic one.

A high density of nodes: The high density of nodes

is due to the way the nodes are positioned (the random mode).

A collaboration among the nodes: The strict constraints of

energy consumption leads sensor nodes to detect
and process the data in a cooperative manner to avoid the
redundant processing of the same observed data, source of
Presentation of Wireless Sensor Networks 18

the loss of energy [4].

1.3.8 The areas of application

The increasingly reduced size of micro-sensors, the cost of

increasingly weak, the wide range of available sensor types
(thermal, optical, vibrations,...) as well as the communication support
wireless cation used, allow sensor networks to invade
several areas of application.
Among them, we can mention:
Military Applications: As in the case of several technologies
Indeed, the military sector was an initial driving force for the devel-
development of sensor networks. The rapid deployment, the
reduced cost, self-organization and fault tolerance of
sensor networks are characteristics that make this type
of networks a valuable tool in such a field.
For example: The detection and collection of information about the
enemy position, monitoring of hostile areas (conta-
mines), the detection of chemical, biological agents...

Environmental Applications: A large number of sensors

can be deployed in a forest or in a conservation environment
Wildlife observation in order to gather various information
on the state of the natural environment and on displacement behaviors
cement.
Among these applications, we find: Fire detection of
rest, precision of agriculture, The tracking of movements of oi-
buckets, of animals and insects...

Medical Applications: One could imagine that in the future, the

monitoring of the vital functions of the human being would be possible
Presentation of Wireless Sensor Networks 19

thanks to micro-sensors that can be swallowed or implanted

tea under the skin. Currently, micro-cameras that can
to be swallowed exist.
For example: Remote control of physiological data
of the human being, the administration of medications inside
hospitals, the monitoring and supervision of doctors and pa-
patients within the hospital...

Commercial Applications: It is possible to integrate cap nodes

participants in the storage and delivery process. The network thus
formed, can be used to know the position, the state and the
direction of a package or a cargo. It then becomes pos-
possible for a client who is waiting for the reception of a package, to have
a real-time delivery notice and to know the position
current of the package. In this field, we find: Detection
and the monitoring of car thefts, interactive museums, The
environmental control in offices and businesses...

1.3.9 Energy consumption in RCSF

The sensors are designed with a limited energy module. Thus,

this last one must be used effectively in order to maximize the du-
life cycle of the network. It should be noted that once the energy of a node is
exhausted, he is considered defective which causes a high probability-
likely to lose network connectivity.

[Link] The main operations due to energy consumption

[Link] energy: Capture energy is dissipated to ac-

perform the following tasks: sampling, processing of si-
signal, analog/digital conversion and activation of the probe
Presentation of Wireless Sensor Networks 20

FIG. 1.10 – The different applications of RCSF [CHE08].

you capture.
In general, the capture energy represents a small percentage
total energy consumed by a node.
b. Processing energy: Processing energy is divided into two
parties: switching energy and leakage energy. The energy
The switching is determined by the supply voltage and
the total capacity switched at the software level (by executing a
software).
On the other hand, the leakage energy corresponds to the energy consumed.
when the computing unit performs no processing. In general,
the processing energy is low compared to that required
for communication.
[Link] energy: Communication energy is de-
cline in two parts: the reception energy and the emitted energy
This energy is determined by the amount of data to
communicate and the transmission distance, as well as by the
physical properties of the radio module. Emission of a signal
Presentation of Wireless Sensor Networks 21

est caractérisée par sa puissance. Quand la puissance d’émission

is high, the signal will have a wide range and the energy consumed
my energy will be higher. Note that the communication energy
represents the largest portion of the energy consumed by
a sensor node.

[Link] Energy consumption model

Heinzelman et al. [HCB00] propose a radio model of consumption

energy animation (Figure 1.11). Thus, the energies required to
issueT x (s, d) and receiveERx(s) messages are given as
suit
To send a message of bits to a receiver far from it.
meters, the transmitter consumes:

ET x (s, d) = ET xelec(s) +ET xamp(s, d)

ET x (s, d) = (Eelec*s) + (Eamp*s*d2 )
To receive a message of s bits, the receiver consumes:
ERx(s) = ERxelec(s)
ERx(s) = Eelec*s
Eelecandamprepresent respectively the transmission energy
electronics and amplification.

FIG. 1.11 - Energy consumption model.

Presentation of Wireless Sensor Networks 22

[Link] The factors involved in energy consumption

Energy consumption depends on several factors which are

explained below:
a. Status of the radio module: The radio module is the component of the
sensor node that consumes the most energy, since it is the one
which ensures communication between the nodes. We distinguish the
four following states:
Sleep state: The radio is turned off.
State transmission: The radio transmits a packet.
State reception: The radio receives a packet.
Idle state: This state causes a loss of energy due to
the useless listening to the transmission channel. To avoid this loss
of energy, a sensor must activate only when necessary, and
the rest of the time he must put himself in a state of sleep.

b. Access to the transmission medium: Since the nodes share

the same transmission medium, the MAC sublayer plays a
important role for coordination between nodes and the mini-
optimization of energy consumption. Indeed, minimizing the
collisions between the nodes help to reduce energy loss.
Thus, the main causes of energy loss are:
The retransmission: The sensor nodes generally have
one single radio antenna and share the same transmission channel
mission. Furthermore, the simultaneous transmission of data
coming from multiple sensors can produce collisions
and thus a loss of the transmitted information. The retransmission-
the loss of packets can result in a significant loss
energy.
Overhearing: The phenomenon of overhearing occurs
produces when a node receives packets that do not belong to it
Presentation of Wireless Sensor Networks 23

destined, As illustrated in the following figure:

FIG. 1.12 - Eavesdropping [MOA08].

Active listening: Active listening (idle listening) of the channel for

a possible reception of a package that will not be received may
causes a significant loss of capacity of the nodes in
energy.
–The overload: Several protocols of the MAC layer
they interact through control message exchange (overhead) for
ensure different functionalities: signaling, connectivity,
establishment of access plan and collision avoidance. All
these messages require additional energy.
Over-emission: The phenomenon of over-emission
(ting) occurs when a sensor node sends the data to
a recipient who is not ready to receive them.
The size of the packets: The size of the messages exchanged in
the network has an effect on the energy consumption of the nodes
emitters and receivers.

[Link] propagation model: The propagation model is

feel an estimate of the average power received from the signal
radio at a given distance from a transmitter. The propagation of
radio signal is generally subject to various phenomena:
reflection, diffraction, and dispersion by various objects.

Data Routing: Routing in sensor networks

is a multi-hop routing. The forwarding of packets from a
Presentation of Wireless Sensor Networks 24

data from a source to a destination is transmitted through several nodes

intermediaries.
In this context, a poor routing policy can have
serious consequences on the lifespan of the network.

1.3.10 Routing in the RCSF

Routing protocols within the RCSF are influenced by

a determining factor namely: The minimization of energy without a
considerable loss of efficiency.
For this, many routing strategies have been created for the
sensor networks. Some are adaptations of strategies that
existed for other types of networks (wireless networks in the sense of the
larger) while others were specifically designed for the
wireless sensor networks.

The proposed routing protocols for RCSF can be

classified according to four ways: According to topology, the establishment
from the road, the communication paradigms and the functioning
of the protocol.
This classification is designed as follows:

[Link] Examples of routing protocols

a)SPIN :
Heinzelman et al. proposed a family of protocols called SPIN
(Sensor Protocols for Information via Negotiation), relying on a
negotiation model to spread information in a network
of sensors.
Communications in SPIN occur in three stages [CHE08]:
When a node wants to send data, it first emits a
message ADV containing a description of the data in question
Presentation of Wireless Sensor Networks 25

FIG. 1.13 - Classification of routing protocols [BEN09].

tion.
A node receiving an ADV message checks its interest base.
If he is interested in this information, he sends a REQ message.
towards his neighbor.
Upon receiving a REQ message, the sender transmits to the interested party
the data in the form of a DATA message.
Presentation of Wireless Sensor Networks 26

FIG. 1.14 - The steps of the SPIN protocol [CHE08].

b) MCFA :
Yeet et al. proposed the MCFA (Minimum Cost Forwarding) algorithm
Algorithm), searching for a minimal path between the source and the sink,
tout en considérant les limites des réseaux de capteurs. Le protocole
aimed at achieving three main goals:
Optimality: By routing data over cost-effective paths
minimum.
Simplicity: Which translates to low energy consumption.
moire, and the non-necessity of identifying the nodes.
Scalability: Given the low memory consumption
and the absence of node identifiers, the protocol can be used
for a large number of nodes. In addition, the construction phase
routes consume only one message per sensor.

Its principle is as follows:

Each node maintains a cost variable, which determines the cost
minimal towards the well on the optimal path. Several measures can
Presentation of Wireless Sensor Networks 27

to be used, depending on the desired application: number of jumps, consumption-

energy management, ... etc.

c) Directed Diffusion :
Directed Diffusion is a data propagation protocol,
allowing the use of multiple paths for routing information.
The well diffuses an interest in the form of a request, in order to inquire.
the network on a particular data. It is based on the pu model-
publish/subscribe. DD relies on four elements: appointment of
data, interest propagation and the establishment of gradients,
data propagation and path strengthening.
DD employs the global flooding of the network. Each node maintains
locally a cache of interest containing the following information:
The description of the interest, using the nominal scheme
When a well requires a data from the network, it propagates
an interest, containing its description as well as the flow of information
desired action.
A set of gradients: When a node receives an interest,
he goes through his cache :
If the cache does not contain any entries related to the received interest,
a new entry is created with a gradient towards the neighbor
transmitter.
Otherwise, the node searches for a gradient towards the
transmitter neighbor, and updates the entry in question accordingly
After processing the cache, the relay node sends the interest to its
neighbors. The simplest method is to use flooding.
Presentation of Wireless Sensor Networks 28

FIG. 1.15 – Steps describing the DirectedDiffusion protocol [CHE08].

1.4 Conclusion

This chapter provided an overview of the field of

wireless sensor networks that are related to the Ad network
hoc.

Wireless sensor networks are of significant interest.

maple and a new step in the evolution of technologies
of information and communication. This new techno-
Logistics generates growing interest due to the diversity of its applications.
sectors: health, environment, industry, and even in the field
athletic.

However, the implementation of sensor networks must

to meet some constraints such as energy consumption
guy, the change in topology, the significant density of the re-
bucket,...etc. These limits have led researchers to suggest some
Presentation of Wireless Sensor Networks 29

monitoring approaches to keep the network operational

as long as possible while maintaining connectivity
in the network (connected network) and a better coverage of the
zone.

As a result, the next chapter will be devoted to surveillance

of RCSF as well as the presentation of the distributed approach that we
have chosen.
Presentation of Wireless Sensor Networks 30
Chapter 2

The surveillance of networks

Wireless Sensors

Surveillance is a device used to analyze the state of the system.

theme. It consists of detecting and classifying failures while ...
serving the state of the system in real time, then diagnosing [GVA07]
to be able to locate the faulty elements and determine the causes.
In general, the design of effective supervision is based on
the combination of surveillance and diagnostic techniques.
As a result, in the RCSF, some sensor nodes can be
stuck or broke down due to a lack of energy, a de-
damaged equipment or an environmental interference. The failure of a
The sensor node should not affect the overall operation of its
network. It is the issue of reliability or fault tolerance.
For the resolution of these problems, several central approaches
proposed lists that are based on diagnosis and reconfiguration
ration at the sink level which causes several disadvantages such as
that: Module overload, Network overload by the mes-
control sages, the transmission times that may be delayed.
defect detection, and the greatest disadvantage is failure
of the sink module.
The monitoring of Wireless Sensor Networks 32

Under these circumstances, this chapter will present a m-e-

decentralized nodes that aim to reconfigure the network in case of
failure based on the principle of redundancy with some
properties of graph theory in order to maintain the connected network
while covering the entire catchment area.
The monitoring of Wireless Sensor Networks 33

2.1 The supervision of RCSF

A supervisor's goal is to ensure the smooth running of a

mission in the presence of problems and manage the best functioning
possible. From this principle, the monitoring of sensor networks
consists of detecting faults and failures then generating solutions
real-time actions to maintain the longevity of the network.
Breakdowns can occur due to a defect at the level of
software or a human error. In the RCSF, the failure is gen-
rarely triggered due to the energy depletion of a node
which is caused by network overload with messages from
controls and even because of the constraints in which the network is
deployed.

2.1.1 The Breakdowns

A failure (or malfunction) of the system occurs when its behavior

statement becomes inconsistent and does not provide the desired result. The
Failure is a consequence of one or more mistakes. A mistake
represents an invalid state of the system due to an error (defect). The
fault is therefore the primary cause of error, the latter causes
the flaw in the system.

[Link] Classification of failures

The following diagram shows a general classification according to the

reason, the cause or the behavior of a failure:

[Link] according to duration:

a1.[Transitory:] Consequence of an environmental impact

temporary, it may eventually disappear without
The monitoring of Wireless Sensor Networks 34

FIG. 2.1 – Classification of failures [CHE08].

cune intervention.
a2.[Intermittent:] A variant of the transient failure, it pro-
occasionally and unpredictably. It is ge-
generally due to the instability of certain characteristics
material or to the execution of the program in a space
particular of the environment.
a3. [Permanent:] continuous and stable over time, the breakdown
permanent persists as long as there is no ex- intervention
turn to eliminate it. A physical change in a com-
causing a permanent hardware failure.

[Link] according to the cause:

b1.[Design failure:] Due to poor structuring of the re-

bucket or a particular component. In practice, this kind
the breakdown should not exist thanks to tests and simulations
actions before the final realization of the network;

b2.[Operational failure:] Who occurs during the function-

The monitoring of Wireless Sensor Networks 35

system development.
It is generally due to physical causes. Furthermore,
we can distinguish, especially for cap networks-
three main causes: Energy, Security and Trans-
mission.

[Link] according to the resulting behavior:

c1.[Accidental failure (Crash):] The component stops.
completely stop working or continue but without re-
turn to a stable state (valid).
c2.[Omission failure:] The component is no longer able to improve
improve its service (total failure).
c3.[Synchronization failure (Timing) :] The component effects-
your treatment but provides the result late.
C4. [Byzantine Breakdown:] This breakdown is of an arbitrary nature; the
the behavior of the component is therefore unpredictable. Due to
very malicious attacks, this type of failure is considered
the most difficult to manage.

2.1.2 Fault-tolerant architectural solution in the RCSF

The solution can be classified into 3 categories:

Battery management: This category is considered as
a preventive approach, where protocols define a dis-
uniform distribution for energy dissipation among the dif-
different sensor nodes; in order to better manage consumption
of energy and the increase in network lifespan. In
Furthermore, the sleep mode mechanism is a management technique.
battery operation. Indeed, the protocols determine deadlines
putting inactive sensor nodes to sleep for better
The monitoring of Wireless Sensor Networks 36

energy conservation

Flow Management: This category includes techniques that

define data transfer management protocols
(routing, selection of transmission channel...etc.). We can-
we will find preventive or curative approaches on the dif-
reference layers (network, data link, etc.) such as:
Multi-path routing: Uses a preventive algorithm to de-
to end several paths from each sensor to the node
collector. This ensures the presence of more than one reliable path
for transmission and offers a quick resumption of the transfer in
failure case on the first selected path (choose one of the
paths that remain).
Road recovery: After detecting a failure, a technique
curative allows the creation of a new, more reliable path for
transmit the data.
Channel allocation: This solution, implemented at the MAC level,
allocate the transmission channel in a way
to reduce interference between neighboring nodes and avoid
collisions during the transfer.
Mobility: Some protocols propose tolerance as a solution.
selection of a set of mobile nodes against failures
charged with moving between the sensors and collecting the data
captured. This will reduce the energy consumed at each
sensor by eliminating its transmission task. A mobile node
is generally equipped with a larger battery than that
of a sensor node.

Data management: The protocols classified in this category

RIE provides better data management and processing.
Two main subcategories are determined:
The monitoring of Wireless Sensor Networks 37

Aggregation: Considered a preventive approach, the operation

aggregation performs an additional processing on the data
raw data captured from the environment. An aggregate node-
It combines data from several nodes into one.
significant information; what considerably reduces the quantity-
data transmitted, requires less energy and increases
thus prolonging the lifespan of the network.
Clustering: One of the important approaches to address the structure
The structure of a sensor network is clustering. It allows for
creation of a virtual backbone that improves the use of resources
rare resources such as bandwidth and energy. Furthermore,
Clustering helps to achieve multiplexing between different clusters.
Furthermore, it improves the performance of the algorithms.
routage. Plusieurs protocoles utilisent cette approche préventive
[FEL07].

2.1.3 Monitoring aspect of a RCSF

[Link] Deployment methodology:

Sensors can generally be placed in an area of interest.

wait, whether deterministic or random. The choice of deployment strategy
depends on the type of sensors, the application, and the environment
operation of the sensor. The controlled deployment of the nodes
is viable, and often necessary when the sensors are expensive, or
when their operation is significantly affected by their position
In certain applications, the random distribution of nodes
is the only possible option. This is particularly true for the
difficult environments such as the battlefield or a region
of catastrophe. According to the distribution of nodes and the level of redundancy
dance, the random deployment of nodes can achieve the objectives
The monitoring of Wireless Sensor Networks 38

of performance.

FIG. 2.2 – Deployment strategies of nodes in a RCSF[YAK].

[Link] Network Connectivity

A wireless sensor network is said to be connected if and only if it

there exists at least one route between each pair of nodes [MER03]. The
connectivity essentially depends on the existence of roads. It is
affected by topology changes caused by mobility or
the failure of the nodes [BEC09].
The principle is that the good coverage will be provided when...ris a
multiple deSr However, if the communication range is limited,
for example, Tr=Sr , connectivity is becoming an important issue,
unless the redundancy in coverage is provided for.

One of the methods used was proposed by "J. Bredin, E. De-

"M. Taghi Hajiaghayi, and D. Rus" in 2005 which is based on the
formulation of K-connected RCSFs [CSS04]. K-connectivity implies
that there are K independent paths between each pair of nodes.
For K > 1, the network can tolerate certain link failures and ga-
rantal a certain capacity for communication between the nodes. The
authors study the problem of node placement to achieve
K-connectivity at the time of network configuration or repair
To create a disconnected network, formulate the problem as a model.
The monitoring of Wireless Sensor Networks 39

of optimization that seeks to minimize the number of supplementary nodes

comments required to maintain K-connectivity.

[Link] The cover

The coverage problem in video sensor networks

is not recent [MSP09]. Several works have already addressed this pro-
problem in several ways. However, there are two categories of
cover:
Coverage of predetermined targets: Involves finding a subset
set of connected nodes that ensures the monitoring of an e-
set of targets whose position is known in advance.
Coverage of an area: involves finding a subset of
connected nodes that monitor the entire area of ...
deployment.

Let there be a network of nodes deployed to monitor a region

of interest, a node is characterized by its coverage radius
its transmission range. The coverage in a sensor network
wireless depends on the density of the nodes. This density is given by
the equation extracted from [CSS04] :

µ(R) = (NΠR2 /A
Where:
The transmission range of a node,
At the calculation level,

The number of nodes located in the air A

[Link] Network Longevity

Extending the lifespan of the network has been the optimization objective.
for most of the communication protocols published for the re-
sensor buckets. The positions of the significant nodes impact on
The monitoring of Wireless Sensor Networks 40

the lifespan of the network. For example, the variations in density of

nodes throughout the region can lead to the des-traffic load
balanced and cause bottlenecks [HWT05].
A wireless sensor network becomes unusable when the connec-
When the activity between its nodes is lost, it is said that the network is dead.

2.1.4 Redundancy

Redundancy, in general, is used to prevent a

malfunction in a system. As in the case of RCFS, the
redundancy is established to extend the lifespan of the network,
all that this last one relates to the excess number of nodes.
Redundancy is divided into two categories:
a) Hardware or physical redundancy.
b) Analytical redundancy.

[Link] Material Redundancy

Physical redundancy involves using several sensors, ac-

tuning devices, processors and software to measure and/or control a
particular variable. A voting principle is applied to the values
redundant to decide whether a fault is present or not. This ap-
close incurs significant costs in instrumentation but proves to be
extremely reliable and easy to implement. It is being implemented
mainly on high-risk systems such as power plants
nuclear or planes.
This approach is applied for RCSF, as it is the means
the easiest way to obtain multiple pieces of information about the same variable
and to have several sensors measuring the magnitude of this
variable.
Physical redundancy suffers from a major disadvantage: its
The monitoring of Wireless Sensor Networks 41

FIG. 2.3 - Hardware redundancy [RIP99].

cost. Doubling the number of sensors amounts to at least doubling the

price of the measuring device. Furthermore, ergonomic constraints
related to the installation of these sensors can limit their use.
The modes that can be applied to the different levels of a
systems are:

Active redundancy: It is suitable for critical systems.

because of its principle that tolerates the fault of one or several
components.

2. Passive redundancy: Based on redundancy using a

single copy, called 'primary copy'; whereas the other copies
are only used in case of failure.

3. Hybrid redundancy: As its name indicates, it is a com-

combination between the two redundancies mentioned earlier.

The choice of the redundancy strategy is made based on the constraints.

and application needs. In the case of sensor networks without
Indeed, active replication is preferable to tolerate failures.
frequent sensor nodes.
The monitoring of Wireless Sensor Networks 42

[Link] Analytical redundancy

A complement to physical redundancy is to exploit the

constraints linking the different variables of the system. These constraints
can often be expressed in the form of analytical relationships
linking known variables (input/output or output/output relationships).
These relationships are called analytical redundancy relations.
The principle of surveillance consists of verifying whether these relationships are
equal to zero (in a statistical sense specified later) using
the measurements taken online on the system. The concept of redon-
analytical dance is based on the use of a mathematical model
of the system to monitor.

Definition 1 the structure of an analytical redundancy relation is the

minimal list of constraints that must be satisfied for this to be the case
also. Each constraint being associated with a component, the structure
An RRA will be rated using the set of corresponding components.
dants. [MAQ03]

The approach using analytical redundancy is broken down into ge-

generally in two distinct phases:
The first concerns the generation of characteristic residues of the
breakdown.
The second step concerns decision-making related to the de-
detection and possibly the localization of a failing element.
She implements break detection techniques and tests.
of hypotheses.
This principle is often difficult to exploit as in the case of the
RCSF due to the frequent change of the related network topology
to battery depletion or even the mobility of the sensors. For
The monitoring of Wireless Sensor Networks 43

for these reasons, hardware redundancy is used especially because of the cost
wireless sensors are becoming smaller and smaller.

2.2 Multi-Agent Systems

A multi-agent system (MAS) is a system composed of an en-

set of agents, situated in a certain environment and interacting
according to certain relationships. An agent is an entity characterized by
the fact that she is, at least partially, autonomous. This can be a
process, a robot, a human being, etc. Long-standing subject of
research in distributed artificial intelligence, multi-agent systems
agents form an interesting type of modeling of societies, and have
to this end broad fields of application, extending to the sciences
humans.
In other words, it is about making a set of entities (agents) cooperate.
equipped with intelligent behavior, coordinating their goals and their
action plans to solve a problem.

2.2.1 Characteristics of Multi-Agent Systems

Each agent has information or problem-solving abilities.

of incomplete problems, so each agent has a point of view
limited;
There is no overall control of the system;
The data is decentralized;
The calculations are asynchronous.
The monitoring of Wireless Sensor Networks 44

2.3 Previous Works

In the field of RCSF(s), several approaches have been proposed.

measures to remedy the problems found following the operation
RCSF(s). In this context, the methods or solutions propose
they are classified into two categories:
Centralized Methods;
Decentralized Methods.
Knowing that researchers are leaning towards the second category
causing centralization problems that can be summarized as:
Overload of the processing module;
– Overload of nodes by control messages.....etc.
As a result, the work concerning the RCSF(s) is more focused on
energy efficiency, service quality and recently towards the
monitoring of RCSF(s). Regarding this last point, we can
quote the following:
In "SELF ORGANIZED SENSOR" [YCC05], the author designed
his monitoring protocol based on the construction of a hierarchical tree
hierarchical using the mathematical model such as the theory of
graphs or a Voronoi diagram that applies to a large
zone due to the use of multi-hop communication as well
The data could be aggregated at the level of intermediate nodes.
diaries.
It is based on the following points:
Collect the positions of the nodes;
Clean the Redundant nodes and detect the boundary nodes;
Detect and track intrusion objects;
Reorganize the nodes when a node dies.
The redundant node is detected according to its coverage area with a
binary image.
The monitoring of Wireless Sensor Networks 45

The EEP Protocol proposed by SAYAD MAYA [SAY09] brings

improvements such as the increase in operating time
network management as well as the reduction of transmission time.
As a result, there is energy savings because of the creation of roads.
is dynamic and is based on a cost function that takes into account
the energy level of the nodes and their distance from the sta-
basic action.

The authors [BAI10] studied monitoring using assistance

multi-criteria decision making by formulating specific requirements that are the
limited resources in terms of load and storage capacities
and calculation, as well as the absence of fixed infrastructure (mobility),
connectivity, and the distance between the sensors face the bad conditions
behaviors that can target it.

In CFIP09 [MSP09], the authors focus on monitoring

videos of Wireless Sensor Networks where they adopt an ap-
close to scheduling the activity of nodes adaptively
video based on the coverage. They headed towards a al-
distributed algorithm where each node organizes its neighbors into a
set of disjoint nodes then based on the activity, each
The node decides whether it remains active or not. The simulation was established based on

of a multi-level model.

Samira ALLAM in [ALL09] is interested in the routing protocol

based on a distributed approach with a multi-agent context
see on self-organization aims to balance a central system
linked to a decentralized system to ensure energy reduction
consumed.
In [BEC09], the author proposed a decentralized model for the
monitoring of a RCSF using a personal simulator designed
The monitoring of Wireless Sensor Networks 46

with C++ for test validation.

After several investigations, we found that the monitoring

RCSF(s) is a field that is expanding, which is attracting researchers' interest.
pressing more and more towards this door that has opened another concept of
search. As a result, we based ourselves on the application of a
distributed approach for Wireless Sensor Networks and si-
Test this approach on an existing open-source simulator.
The distributed or Multi-Agent approach for:
Adaptation to reality;
Cooperation;
The resolution of complex problems;
Effectiveness;
Reliability.
The surveillance of Wireless Sensor Networks 47

2.4 Description of the decentralized approach

sée)

L’approche suivie consiste à la définition d’un réseau de capteur

wireless with the principle of agents: different roles assigned to nodes
to form groups (areas or clusters), the same algorithm is
used for network reorganization in case of failure when using
without the principle of redundancy with graph theory.

2.4.1 Network Topology

The wireless sensor network is deployed randomly in a

zone, these nodes follow a dynamic topology, i.e. the cap-
ters may move or even the addition or disappearance of caps.
payers may be considered.
The network is divided into groups of sensors where each node has a
role assigned based on its neighbors during the re-organization
bucket. This principle is followed to make a set of entities cooperate.
equipped with behavior, coordinate their goals and their action plans
solutions to solve a problem [LEB].

2.4.2 Definition of roles

Roles can be defined as follows:

a) Representative Node: It is the node that administers the com-
communications within the sensor group. It assumes all the
requests to send messages from members of their region. It is
the node that is responsible for the group's communication with
the outside. It should be noted that a representative can never have a
neighbor of the same role.
The monitoring of Wireless Sensor Networks 48

b) Liaison Node: Allows the representative nodes of communi-

wants (he is on the borders of several sensor groups). He
is responsible for handling communications between the regions.

c) Simple Node: It is nothing more than a simple sensor, it captures

the requested information and communicate it to the Sink via its
representative.

d) Sink Node: It is the main station that can communicate.

with all the nodes to collect the desired information.

FIG. 2.4 - Organization of sensor nodes by regions [ALL09].

The monitoring of Wireless Sensor Networks 49

[Link] Role Assignment Algorithm

The following algorithm explains the role assignment procedure.

Role Assignment
{
if(NombreVoisin !=0)
{
if(NumberOfNeighborRepresentatives == 0)
No neighboring Representative, so the sensor becomes one.
RoleAffecté = Representant ;
else
{
if(NbreVoisinRepresentant == 1)& &(RoleAffecté != Representant)
{
there is only one representative linked to the nodes
//which forces other sensors to become simple nodes
RoleAffecté = Simple ;
}
else
{
several representatives around the node then this
sensor can only be a link node
RoleAffecté = Liaison ;
}
}
}
}
The application of such an algorithm leads the network to represent itself as

the following way :

The monitoring of Wireless Sensor Networks 50

FIG. 2.5 - Role assignment in a CRSF.

2.5 The adopted routing protocol

The following routing protocol is based on the principle of assigning

role playing which consists of several steps summarized in the
following points:
Neighborhood detection;
Change of roles;
Detection of neighboring groups;
Conflict resolution between representatives;
Management of redundancy.
Fault detection mechanism.
Knowing that the nodes of this system interact by sending
messages. The interactions between the 3 agents allow for understanding
the functioning of this protocol. These interactions are defined by
with respect to the assigned role.
The monitoring of Wireless Sensor Networks 51

2.5.1 UML Diagrams

The diagrams following the UML method, more precisely the

State-Transition Diagram and Sequence Diagram are developed
to present the different states of a node and to give a
a general and clearer idea of the principle of the approach taken.

FIG. 2.6 – State Transition Diagram.

The surveillance of Wireless Sensor Networks 52

FIG. 2.7 - Sequence Diagram.

[Link] Neighborhood detection

The neighborhood of a node is detected to ensure the connection

of the network. In this context, each sensor broadcasts a message called
Message HELLO (to all nodes in a single hop). This message is
defined as follows:

Type Source Destination NSaut

The monitoring of Wireless Sensor Networks 53

Each node receiving this message (HELLO) responds with a message

message for destination (non-redistributable) structured as follows:

Type Source Destination Rôle Groupe Position NSaut

[BEC09]

By sending these messages, all the nodes can update.

their neighborhood table and from this principle the change of role can
affected the sensor nodes.

2.5.2 Change of roles:

During the detection of change in the neighbor table by

a node. The latter executes the role change algorithm ex-
mentioned earlier by sending a "ChangeRole" message that has
the structure presented below. Knowing that each node receives
this message will force the execution of the role assignment algorithm
while modifying its neighbor table.

Type Source Destination Rôle Groupe [BEC09]

FIG. 2.8 – Introduction of a new node.

The surveillance of Wireless Sensor Networks 54

2.5.3 Detection of Neighboring Groups

Representative nodes are defined with a table containing

the neighboring groups. The latter must be updated according to the
knowledge of the links between the groups using the
message HELLO.
This figure presents the operation mentioned on group detection.
adjacents.

FIG. 2.9 – Detection of adjacent groups.

2.5.4 Verification of group consistency:

The inconsistency problem can be detected by a simple node.

Because this problem can occur if and only if there are two
disjoint groups i.e. the simple nodes can see each other but their representatives
sentences cannot communicate (Lack of a connection node).
The following diagram illustrates this problem:
At the moment a node detects the inconsistency, it propagates a mes-
said "VérifCohérence" which is formed of :

Type Source Destination Rôle Groupe GrpVoisin [BEC09]

The reception of the message 'VerifInconhérence' triggers the process-

hard role assignment. After executing this procedure and
If the inconsistency is confirmed, a Representative node may change.
his role by assigning to another node the role of Representative in it
The monitoring of Wireless Sensor Networks 55

FIG. 2.10 – Correction of the inconsistency problem.

seeing the message 'PassRepresentative' as shown by the structure

next :

Type Source Destination Groupe [BEC09]

2.5.5 Conflict resolution between representatives

A sensor network follows a dynamic topology, i.e. the to-

Apology can change due to the mobility of a node (exhaustion)
energy....). This configuration can produce a conflict between the
groups in the case where two representatives become neighbors.

This problem can be solved using the message 'ResConflit' which

in the following format:

Type Source Destination Score [BEC09]

Knowing that the 'Score' defines the remaining energy of each Repre-
to be able to designate the best node in terms of energy.
As a result, the Representative node that receives this message can either remain
Representing if its score is high by sending the message 'ResCon-'
"flit", or become a Simple node by broadcasting the message "Change-
Role.
The monitoring of Wireless Sensor Networks 56

FIG. 2.11 – Conflict resolution between Representatives.

2.5.6 Redundancy Management

A node is said to be redundant if and only if there exists another

node having the same role capable of ensuring precisely the same
tasks of the candidate node.
This redundancy management mechanism is used to maintain
a long functioning of life on the network with a minimum
of nodes collaborating to be able to wake up the other nodes in case
of a sensor failure.
This process is applied to all nodes whose roles are either
Simple or Connection in the following manner:

[Link]
A Simple node "S1" is said to be redundant if there exists another node
Simple 'S2' belonging to the same group as S1 such that:

distance(S2,R) >= distance(S1,R)

R: Representative Node
d : Distance.

This principle can be formulated using Ga's property.

briel :
The monitoring of Wireless Sensor Networks 57

Redundant(S1)/S1_GiThere exists S2 in Gi etS1 Circle(S2, Ri ] [BEC09]

FIG. 2.12 – Redundancy of simple nodes.

[Link]:
An L1 link node is considered redundant if and only if it
there is another L2 linkage node such as the representatives Ri
couvert par le nœud L1 sont inclus ou égal à l’ensemble des
nodes representing covered by L2.
We can phrase it this way:

Redundant(L1)/L1GIThere exists L2 in Gi /Grp(L1)⊆Grp(L2)[BEC09]

-Gia group of nodes having Rias a representative.

-Lja link node in the groupGi .
- Grp(Lj the set of adjacent groups accessible by RIà
through the node liaisonLj .
The Redundant Predicate (Lj valuing 'True' when the nodeLjest
declared redundant.

[Link] The principle followed to manage redundancy:

The redundancy of the nodes is managed by the exchange of messages.

Knowing that only the Representative has the ability to change the state
of a node that is part of its group based on a message
ChangeState
The monitoring of Wireless Sensor Networks 58

or to put them to sleep.

The composition of the message is as follows:

Type Source Destination NS [BEC09]

Knowing that:

-NS"NEWSTATE"isthenewstateintendedforanodebyits
representative.
The mechanism is presented in (Figure 2.13):

FIG. 2.13 – State change of a node.

[Link] Fault Detection Mechanism

The fault detection mechanism is utilized at the level of

this approach. In this context, the periodic exchange of the message
"HELLO" allows the Representative node to discover the failure.
of other nodes belonging to its group.
The principle can be described in these points:
The Representative node periodically broadcasts the message 'HELLO'
to the active nodes.
All nodes receiving this message respond with the message "HEL-"
LORep" i.e. that these nodes always collaborate in the network.
The node that does not respond is considered faulty by the Re-
presenting, the latter wakes up all the neighboring nodes and launches the
The monitoring of Wireless Sensor Networks 59

Role assignment process by spreading the message 'Chan-

geEtat

2.5.7 Information Collection

A sensor network is configured for information collection.

tion, such as temperature, pressure, or multi-information
average.
The approach used is based on the use of two simple messages.
of scalar type Data and DataReq defined as follows:

Type Source Destination Mesure

Type Source Destination

The sink broadcasts its request using DataReq to all nodes.

The latter exploits this message according to their roles.

Link Node: Passes the request to all its neighboring representatives.

Simple Node: Sends the measured value to its representative.

Node Representative: Responds to the request by sending the info-

information captured by its group using the optimal path that
is known for the periodic sending of the message 'HELLO' with the
champ "HC" (Hop Count).
The monitoring of Wireless Sensor Networks 60

2.6 Conclusion

This chapter presents the protocol used and proposed by [BEC09]

for redundancy management and fault tolerance.

The necessity of the long duration of a wireless sensor network

led researchers to propose different consumption mechanisms
Energy management, monitoring at the level of these networks. As a result, the
group configuration allows for the self-organization of nodes without
call on the sink.

The random deployment of nodes involves the presence of the cap-

redundant actors, the latter are put on standby in order to minimize
the energy consumption of the nodes knowing that these sensors
are awakened if needed.

In the following chapters, we will present the simulation tools.

existing relations while justifying our choice, thus we will show the
simulation results to validate the studied approach and propose
other suggestions.
The monitoring of Wireless Sensor Networks 61
Chapter 3

Simulation of monitoring in RCSF

Information technologies are increasingly spreading which

leads to the emergence of several systems and thus different me-
management methods.
In the network model, the study mainly involves studying
its atomic components. This principle is necessarily applied to
wireless sensor networks, the various proposals from researchers
cheers to remedy the problems of the RCSF such as consumption
energy production, the network operating duration, the constraints
information collection... requires a testing phase before implementation
in place. For this, the most reliable and least expensive solution
It consists of "The Simulation".

The simulation of sensor networks mainly consists of

the reproduction of the behavior and functioning of nodes
sensors in a computing environment; for reasons such as:
The repetition of experience, The addressing of complex systems, The
time savings and the variation of simulation parameters while
the actual simulation proves to be costly, if not impossible in some
case

In the following, we will briefly discuss simulators of

Simulation of monitoring in RCSF 63

existing networks, of the simulator chosen for our simulations "OM-

NET++" and the sensor network simulation platform
wireless 'CASTALIA'.
Simulation of monitoring in RCSF 64

3.1 General Information on Simulation

In our days, simulation is experiencing considerable growth, and this

thanks to the interest presented by the computational models of systems
simulated;

Definition 2 Simulation consists of the computer modeling of a

any system, by providing a representation of all the entities of this
system, their own behaviors, as well as their interactions. It highlights
made available to the user an experimental environment that
we can vary the parameters.

Thanks to the advances made in the field of development and

programming techniques, we now have
very powerful programming languages. Thus, it becomes possible
to create a simulator in a programming environment
existent.

3.2 Types of simulation

Depending on the type of events in the simulation, we distinguish

we have two types of simulation systems: discrete systems
and continuous systems.

3.2.1 Discrete simulation systems:

Are systems for which the variables concerned by the if-

The modulation changes state only at a finite number of points on the axis.
time. These systems are also called: simulation systems.
Simulation of surveillance in RCSF 65

discrete event simulation.

3.2.2 Continuous simulation systems:

These are systems for which the variables can change.

state at any moment during the simulation.

3.3 The existing network simulators

There are several network simulators such as: NS2, OMNET++

OPNET, GLOMOSIM, JSIM...etc. [TBL08][CET05]
Among them, we will mention a few simulators like NS2, GLOMO-
SIM and of course our OMNET++ simulator.

3.3.1 NS2 :

NS is a widely used discrete event simulator in the

field of research for anything related to networks.
The use of the term 'NS2' specifies the version of the simulator.
NS.
It serves as an important support for the simulation of protocols.
TCP, routing protocols, multicast protocols.
The development of NS follows an object-oriented approach that uses
two programming languages: C++ and TCL. The basic modules
the simulator and the protocols are implemented in C++ with a
TCL layer on top, which provides a flexible and easy-to-use interface.
use [5].
The following figure illustrates the different internal components of the si-
simulator:
Simulation of monitoring in RCSF 66

FIG. 3.1 – Architectural description of the NS2 simulator.

3.3.2 GloMoSim

GloMoSim (Global Mobile information system Simulator) was

designed according to a similar 'layered' architecture
of the seven OSI layers for networks [MAK08].
It was developed at the UCLA Parallel Computing Laboratory.
story using the PARSEC language.
PARSEC (PARallel Simulation Environment for Complex systems)
is a programming language derived from the MAISIE language. All
two are intended for discrete event simulation. They have been
developed from the C language at UCLA (University
California Los Angeles.
Simulation of surveillance in RCSF 67

The packet transfer in GloMoSim is done according to the following scheme.

to want

FIG. 3.2 - Packet transfer in GloMoSim.

Simulation of surveillance in RCSF 68

3.3.3 OMNET++

OMNeT++ IDE (Integrated Development Environment) is based

on the Eclipse platform. It is an open source environment that
provides tools for the creation and configuration of models
networks (the NED and INI files) and tools for the execution of a
set of programs as well as for the analysis of simulation results
tion [MSK05].
OMNeT++ seems to be the best among the open source solutions and
freeware. OMNeT++ seems to attract the community more and more
scientific and an increasing number of models are available.
In the following, there will be a more detailed view of the simulator as well.
the reasons for choosing this simulator.

3.4 Comparison between the simulators

Simulation is often cheaper than experimentation and

is much less risky when man is part of the
studied system. The results can be obtained much more
quickly. The simulation (especially numerical) is based on a
knowledge of phenomena that can only be obtained through ex-
experimentation.
A simulation can therefore only be conducted if one has an ac-
who has sufficient knowledge gained through experimentation
about previous and similar phenomena. Whatever the qua-
limited in the simulation, it does not fully replace the experiment.
Here is a general table describing the advantages and disadvantages
nients des simulateurs décrit auparavant [RSA09] :
Simulation of monitoring in RCSF 69

Simulators Advantages Disadvantages

NS2 Use Object-Oriented Designed for networks
filaires,
◦Allows simulation of ◦Difficulty in adding
behaviors of proto-new models due to
coles standard dependencies between
modules,
Numerous protocols Difficult integration to
available, other applications,
◦ Simulator Multi-◦The low performance
Couches. network simulations
important,
Simulation scenario
describe in Otcl.
GloMoSim◦Initially designed for ◦Need to master
wireless networks (Especially the Parsec for all
Ad Hoc networks, personalization around the
core
◦Execution of the simula-◦Installation quite com-
relatively quick actions, plexus.
Simulation environment
parallelizable computation.
OMNET++ ◦ Modular architecture ◦ Does not support
allowing the integration of the case of cap networks
new models, tears.
◦Use of C++ (and re- ◦Few models for the
C# implementation for wireless networks,
development of the core
Base classes of Model Description
Simulators can be in NED language.
extents and personali-
sights,
Model design
Simulation of surveillance in RCSF 70

3.5 The OMNET++ Simulator (Objective Modular Net-

workTestbed in C++)

3.5.1 Choice of the OMNET++ simulator

The deployment of a sensor network requires a simulation step.

testing before its installation on site. The simulation allows for testing
at a lower cost the performance of a solution.
OMNeT++ is an event-driven simulation environment.
secrets based on the C++ language, an open-source application and under
GNU license [IDE04]. It is fully programmable, customizable.
and modular thus thanks to its flexible and generic architecture, it has
has been successfully used in various fields, including [OMN04]
[?] :

Queueing network modeling,

The modeling of communication protocols,
The validation of hardware architectures,
Performance evaluation for software systems
plexes.

OMNET++ will be our simulation environment, thanks to its ar-

modular architecture, it will involve extending the simulator by implementing
such a new model specific to sensor networks.

3.5.2 Presentation of OMNET++

OMNET++ is an event-driven simulation environment.

Secrets. Used for the simulation of communication networks, and
other distributed systems [ABD09].
Thanks to its modular architecture, OMNET++ is very widely re-
pandu in various fields of applications such as:
Simulation of surveillance in RCSF 71

FIG. 3.3 – The launch of the Omnet++ simulator.

Modeling of communication protocols,

The modeling of wired and wireless networks,
The modeling of distributed systems,
The HardWare architectures,
In general, it can be used for any system.
discrete events that can be modeled according to entities com-
communicating by sending messages.

OMNET++ is based on the Eclipse platform. [?] It provides some

tips for the creation and configuration of network models (the
NED and INI files) and tools for batch execution of pro-
grams as well as for the analysis of simulation results.

3.5.3 Architectural description of OMNET++

The OMNET++ models consist of a set of modules

hierarchically nested as shown in (Figure 3.4):
Simulation of surveillance in RCSF 72

The Simple modules are written in C++ using the library of

FIG. 3.4 - Modular architecture of the Omnet++ simulator.

OMNET++ simulation, These contain algorithms

related to the implemented model. The grouping of simple modules
consists of composite modules knowing their communications
are managed through connections between the modules via 'gates
(ports) ".
At a higher level, the system module is created by the user.
It is a special module that has no connections with the environment.
external environment, but rather with its internal components (Modules
simple and compound) [IDE04].
Modules can assign themselves parameters assigned to the mo-
modules in the network description files (NED files) or
again in the configuration file "[Link]". These parameters-
meters are useful for customizing the behavior of
simple modules or for the configuration of the topology of
model.

3.5.4 Installation of the OMNET++ simulator

The installation of OMNET++ is done in different stages following a

installation procedure described in the downloaded package according to the
installed operating system [OMN10]. The elements installed on
Simulation of monitoring in CRSF 73

the computer will be as follows:

An internal simulation library.

A compiler for the NED (NED compiler) descriptive language of topology.
A graphic network editor for NED files (GNED).
An Omnet++ executable.
A graphical simulation IDE interface.
A model documentation tool (opp_neddoc).
Other utilities (makefile creation tool, etc.).
Documentation, sample simulations, etc.

3.5.5 The main files of OMNET++

The different files are [OMN04]:

[Link] File (.Ned):

Use the NED language for network description. It can be used

read in 2 modes: Graphic Mode or Text Mode which allow to
describe the parameters and ports of the module. The errors committed
are indicated in real time by a red point located to the left of
code.
An example of a Ned file in 'Source' & 'Graphical' mode is
presented in (Figure 3.5) and (Figure 3.6).
Simulation of monitoring in RCSF 74

FIG. 3.5 – NED file in graphical mode.

FIG. 3.6 – NED file in text mode.

Simulation of monitoring in RCSF 75

[Link] File (.ini):

Is closely linked to the NED file. Allows the user to ini-

adjusted the parameters of the different modules as well as the topology of the re-
bucket.
Hereisanexamplepresentedbelow:

FIG. 3.7 – Example of a *.Ini File.

[Link] File (.msg) :

The modules communicate by exchanging messages. These deriv-

Invoices can be declared in a file with the extension (.msg)
where data fields can be added. OMNeT++ will translate the
message definitions in C++ classes.
The following diagram can provide a more detailed idea about the
execution development of a simulation in Omnet.
Simulation of monitoring in RCSF 76

FIG. 3.8 - Execution of a simulation under OMNeT++.[?]

3.6 The OMNET++ platforms

The OMNeT++ simulator is not specialized for networks of

wireless sensors, For this there are several extensions, platform
and simulators based on OMNET++ that try to address this gap
like 'Mobility FrameWork', 'Mixim', 'Castalia', etc.

3.6.1 Mobility FrameWork

This platform ensures the mobility of hosts in the network,

connectivity thus a wireless channel. It supports the simulation of
wireless networks, mobile networks, and even sensor networks
wireless [LWI07]. For your information, Mobility Framework 2005 was the
preferred platform for Ad Hoc network simulation.

It is made with basic modules, so the user can

the used in order to create and implement new protocols
for the simulation.
Simulation of surveillance in RCSF 77

The core of Mobility FrameWork consists of [DSR03]:

Mobility Controller (Mobility Management).
Channel Control (The dynamic management of connection between the
hosts.

FIG. 3.9 – Mobility and Channel Control Architecture.

The simulation of a network with such a platform is based on the

understanding of the principle and operating properties of
Mobility Framework (detailed below).

[Link] The structure of a mobile host

The internal structure of a node (Host) in Mobility Framework is

schematized in the following figure [LWI07]:
Simulation of monitoring in RCSF 78

FIG. 3.10 – Internal architecture of a node in MF.

3.6.2 Mixim

Mixim [ABD09] is a simulator that incorporates and develops plu-

existing frameworks for wireless and mobile simulations in
OMNeT++. Although OMNeT++ does not have clear support from the mo-
simulation models for wireless communication, Mixim provides
detailed models of wireless channels (fading, etc.), the connec-
wireless activity, mobility, obstacles and MAC protocols such as
IEEE 802.11b and IEEE 802.15.4. Mixim offers modules for a
easy implementation of new MAC protocols like Time Di-
vision Multiple Access (TDMA) or based on hybrid protocols
brides (e.g. Z-MAC). Mixim is written in C++ language like
Castalia and OMNeT++.

3.6.3 Castalia

Castalia [ABD09] [LIL09] is also a level simulator

for sensor networks based on OMNeT++. It is a simu-
generic lathe with a wireless channel and a radio-based model
Simulation of surveillance in RCSF 79

measured data. As illustrated in (Figure 3.11), Cas-

Talia defines three main modules: the node, Physical Processes
and wireless channels. Castalia is developed with C++.

FIG. 3.11 – The connections of the modules under Castalia.

The node module is actually a composite. (Figure 3.12) shows

the internal structure of the node module. Solid arrows mean the
Message passage and the dotted arrows indicate an interface
between modules and function calls.
For example, most modules call the manager of
resources to report energy consumption. The Module of
the application is the most affected by the user, most often by the
creation of a new module to implement a new al-
algorithm. The MAC modules, Routing, as well as the mod-
ability, are also modified, generally the creation of a new-
calf module is involved in the creation of a new protocol.
All existing modules are very much agreed upon by many...
meters.

Note 1 According to the description of the 3 previous platforms, one

one can easily conclude that the "Castalia" platform is the most suitable.
Simulation of surveillance in RCSF 80

FIG. 3.12 - The main components of a node.

prayers for our project. Its framework of 'Wsn' facilitates the processing of
our application as well as the evaluation of the new protocol.
In addition, Castalia is based on OMNeT++, one of the most popular simulators.
pulses for wireless sensor networks.

3.7 Details about Castalia

As previously described, Castalia is a platform.

under OMNeT++ designed specifically for RCSF [SAY09].
The structure of Castalia [BOU10] is also reflected in the hierarchy-
Head with a source code. Each module contains a "ned" folder.
who defines the module itself. If the module is composed, the sub-
existing directories define the sub-modules with a code of
C++ (cc, header files) which defines its behavior. This hierarchy-
The complete folder 'ned' defines the overall structure of the simulation.
the caregiver of Castalia.

The main modules of Castalia are:

Simulation of surveillance in RCSF 81

3.7.1 The MAC module

The MAC layer plays a very important role in infecting the diff-
rents states to the nodes to improve energy efficiency: node
in transmission (Transmit), listening (Idle), or in sleep (Sleep).
The 'Radio' module modifies these states based on specific cases.
Users can modify the settings in the illustrated examples.
very at the Castalia level for the MAC layer using the file
of the OMNeT++ configuration (.ini). By simply adjusting certain
parameters of the MAC files 'CSMA/CA', 'TMAC', and 'AMCC' of
different simulations will be managed.

3.7.2 The Radio module

It is designed based on wireless sensor networks with three

states: sleep, transmission, and listening. In addition, consumption-
Energy production will be different in each state. It is the characteristic-
The main topic of this module. The state diagram can provide a
general idea on the state change.

FIG. 3.13 – The finite state machine of the Radio module.

Simulation of surveillance in RCSF 82

3.7.3 Wireless channel

The wireless channel module is designed to model the environment

wireless, which is a dynamic evolution of the environment
for the static and mobile node. There are four main models of
wireless channels that are designed:
1) the average modeling of route loss;
2) the modeling of temporal variations;
3) the interferences;
4) simple modeling.
Castalia uses certain formulas to estimate each of the four
models based on many realistic experiences.

3.7.4 The Routing module

The most important function of the network module is to receive

the MAC module message, and send a message on request. He
depends on the header information, there are two types of algorithms
routing in Castalia: simpleTreeRouting and multipathRingsRou-
ting.
All routing modules share 3 parameters:
1)maxNetFrameSize :détermine la taille maximum de paquet.
2) the netDataFrameOverhead: places additional overhead costs
to application packages,
3) netBufferSize indicates the size of the buffer found in the mo-
dule.

3.7.5 The Application Module

The Application module is the main module used for controlling

read other modules such as the resource management module,
physical processing module, routing module, and module
Simulation of monitoring in RCSF 83

radio. Thus this module also uses the message mechanism

to perform different actions.

3.8 Simulation Tools

The installation of the OMNET++ simulator and the Castalia platform.

is as follows:

3.8.1 Installation of OMNET++ Version 4

Before starting the installation, you must first download

download the omnetpp4 source code from the site ([Link]
not forgetting to check the Linux version, then copy the source into
le dossier voulue et se positionner avec l’invite de commande dans
this file and follow the following steps:
1. sudo tar zxf [Link] -C /usr/local/ ;
2. cd /usr/local ;
3. sudo ln -s omnetpp-4.0b2 omnetpp ;
4. sudo apt-get install bison, flex, blt, lmodern, giftrans, doxygen,
libxml2-dev, graphviz, imagemagick ;
5. sudo apt-get install tcl8.4, tk8.4, tcl8.4-dev, tk8.4-dev ;
6. export LD_LIBRARY_PATH=$LD_LIBRARY_PATH :
/usr/local/omnetpp/lib ;
7. export TCL_LIBRARY=/usr/share/tcltk/tcl8.4 ;
8. export PATH=$PATH:/usr/local/omnetpp/bin;
9. cd /usr/local/omnetpp ;
10. sudo ./configure ;
11. sudo make .

Note 2 Steps 6, 7, 8 must be added to the file "∼/.bashrc"

for future convenience.
Simulation of surveillance in RCSF 84

3.8.2 Installation of Castalia3.1

The steps are as follows [BOU10]:

1. Decompression of the Castalia archive already downloaded from the
site ([Link] with
the order:

xzf [Link] ;

2. Importing the project into OMNeT++ IDE: After starting

In Omnet++, importing the project is done with a right click on
General->Exiting project into workspace and choosing the root
of the project.

3. Execution of an example: Once the import is complete,

a right click on the file [Link] in any example
launch the simulation and generate a *.txt file that is defined as
the trace file of the simulation.

[Link] The Castalia Commands

Must be located in the desired directory:

1.../../bin/Castalia Displays the list of files and configurations.
2.../../bin/Castalia -c GeneralStart the execution of the confi-
configuration. Which results in a trace file.

3. CastaliaResults Displays the results obtained during the execution.

4. CastaliaPlot displays the graph of the simulations.

3.9 Implementation of a network

Castalia has considered the development of new

protocols within this platform. In this context, the files
template facilitates the creation of a new network with the different
Simulation of surveillance in RCSF 85

couches.

For the implementation, it is enough to copy the necessary files.

in a new directory of your choice.

We have adopted this principle and created a new directory.

pelé "RCSF" ensuite copier les fichiers"[Link]", "yourAppli-
"[Link]", "[Link]", "[Link]". These contain
roots of procedures, where everyone can modify and elaborate their
protocol.

The program will only be complete with the addition and programming.
files '.cc', '.h', '.msg' of the different modules.

Once all the files are complete, the Castalia platform

will take this new network into consideration only by executing
the following compilation steps:
1. Create the Makefile by executing "opp_makemake -f"
Compile using 'make'.
After these steps, and only if there are no errors, a new
the "youNetwork" file will be created by executing the simulation.
to cut
Simulation of monitoring in RCSF 86

3.10 Conclusion

Dans ce chapitre nous avons présentés les différents simulateurs

existing for wireless sensor networks. Among these simulators,
our choice was fixed on OMNET++ mainly because of its
modular construction and its flexibility.

The lack of the OMNET++ simulator at the level of protocols

Sensor networks have been solved with the implementation of the differe
flat-rate platforms such as: Castalia.

In the following chapter, we will describe our programming steps.

of the protocol proposed in chapter 2, the most integrated within the
OMNeT++ and Castalia simulator at the different layers
thus we will analyze the simulation results of a network with and without
redundancy.
Simulation of surveillance in RCSF 87
Chapter 4

Implementation and Simulation

The evolution of computing has been marked by the emergence of

miniaturized microelectromechanical systems: sensor networks
wireless sensors. Many new applications are emerging each
day of these systems, but there are still many challenges to overcome.

To solve the problems posed by network design

of sensors, it is imperative to go through the simulation of these re-
buckets. However, such networks are never fully taken
in charge of existing network simulators.

The necessity of going through simulation has already been demonstrated pre-
recently. Our study on some network simulators has
leads to choosing a particular simulator: OMNET++.
It was necessary to weigh the advantages and disadvantages presented by each.
simulators and make compromises in this regard.

Our work consists, fundamentally, of adapting a re-

buckets (OMNET++) with the use of Castalia in the particular case of
sensor networks.

In this chapter, there will be a detailed presentation on the implementation.

Implementation and Simulation 89

improvement of our monitoring method as well as the results obtained

points that will be discussed in this same chapter.
Implementation and Simulation 90

4.1 Work Environment

We will detail the tools used in the realization of our

simulation.

4.1.1 Hardware Environment

The simulation was carried out on an Acer computer.

configuration is:

Processor Core2Duo 2.0 GHz

Memory 2GB DDR2
Hard disk 160 GO
2 GB Memory Card

4.1.2 Software environment

Our simulation was conducted in the software environment ...

vant :
Operating System: Linux Distribution Ubuntu.
The OMNet++ 4.0 simulator;
– The Castalia Simulator 3.1.

4.2 Simulation Process

Here is the simulation process we have adopted for the e-la-

boration of our project summarized in (Figure 4.1)
Implementation and Simulation 91

FIG. 4.1 - Simulation Process.

4.3 Project Implementation

4.3.1 Objectif

Our objective is to add at the Castalia simulator level a

new approach and thus test the validity of the latter.
Following this, we found ourselves faced with implementing complete
set a new module in the network layer and add
new concepts at the level of other layers to be able to
to achieve our goal.
For this, we first created a "node" module with everything
its sub-modules, then implement the various utility messages.
those and then program the Network layer module and finally
simulate our approach.

4.3.2 Description of the design of a node

A wireless sensor network is characterized by deployment

random nodes.
Implementation and Simulation 92

For our simulation, the first thing we had to model is the mo-
dual node with its different sub-modules (The protocol stack).

For a better explanatory view, the graphic file ".Ned" illustrates

the different modules used.

moduleWsnNode
{
parameters :
doublenumHosts ;
@display(bgb=250,250,white ;bgp=10,10 ;i=misc/node_vs,gold)
gates:
inputradio@directIn;
submodules :The modules that make up the node
battery : Battery@display("p=200,65,i=block/bucket") ;
basicMobility : BasicMobility
{
@display(p=206,174
}
Nic {
@display(p=94,190 ;i=device/card") ;
}
wsnNetwLayer : WsnNetwLayer {
@display(p=103,117" ) ;
}
wsnApplLayer : WsnApplLayer {
@display(p=94.41
}
connections :
}
Implementation and Simulation 93

FIG. 4.2 – Graphical file '.ned' of the node.

According to the 2 files, we can summarize that the node in our si-
the simulation was designed so that it contains the different layers
described by the sub-modules as follows (Top to Bottom):

Real Layer under Module describing the layer

Application Layer WsnApplLayer
Network Layer WsnNetwLayer
Phy Layer + MAC Nic

In addition to these submodules, there are the following submodules:

"Battery": un réseau de capteur est nécéssairement modélisé par

a battery at the level of each sensor.
"mobility": Pour pouvoir définir la topologie du réseau ainsi pour
knowing the position of each sensor.

In the following sections, we will detail the programming of each.

couches with their code '.cc' and their file '.ned' knowing that these
Implementation and Simulation 94

last ones are implemented from the basic modules of Casta-

lia.

4.3.3 Network Description

The network structure is defined in the file ([Link]).

The latter, as mentioned before, consists of two parts: The
Graphical part and the Source part (Figure 4.3).
A wireless sensor network consists of the described node module.
previously thus a channel for communication between node
and physical process for the management of data captured by the
sensors.

The connections between nodes are managed at the level of this structure.
true in "Connections" but we chose to leave a ge-
random generation to get closer to the concepts of Networks
Wireless Sensors.
Implementation and Simulation 95

moduleWsnNetwork
{
parameters :
doubleplaygroundSizeX ;
doubleplaygroundSizeY ;
doublenumNodes ;
@display(bgb=playgroundSizeX,playgroundSizeY,white; bgp=0,0
submodules :
channelControl : ChannelControl
{
parameters :
playgroundSizeX = playgroundSizeX ;
playgroundSizeY = playgroundSizeY ;
@display(p=33,27
}
node[numNodes] : WsnNode ;
{
@display(p=201, 47") ;
}
connections allowunconnected :
All connections and gates are to be generated dynamically.
}

FIG. 4.3 – Graphic file '.ned' of the network.

Implementation and Simulation 96

4.4 Implementation

We will start to give an overview of the messages.

supplemented, then enter into the heart of the source codes implemented
tea at the level of the different layers.

4.4.1 Structure of the messages used

[Link] At the Application layer

The messages used are simple messages because our

The approach is more concerned with the other layers.
In OMNeT++, messages can be declared in a file.
.msg. The structure of the message 'DATA' used at this level
the layer is as follows:

FIG. 4.4 – " .msg " file of the Data message.

We tested our approach with an existing module under

Castalia, we have just modified the routing of the message from
data that occurs between Simple nodes and Representative node.

[Link] At the Network layer

It is this layer that interests us the most, Consequently we com-

to illustrate the different messages implemented within
the Network layer.
The first message sent by the nodes is the message
Hello. This allows the nodes to know their neighbors and
Implementation and Simulation 97

to fill the neighbors' table.

FIG. 4.5 - The structure of the message Hello and HelloRep.

The HelloRep message has the same fields as Hello and more.
the 3 fields described.
Role assignment message, Conflict resolution (Figure 4.6)
Verification of consistency and Change of role (Figure 4.7) generated
under Castalia (OMNeT++).

FIG. 4.6 – The structure of the message 'Pass Representative' and 'Conflict Resolution'.
Implementation and Simulation 98

FIG. 4.7 - The structure of the message Verification of consistency and Change of
role.

4.4.2 Overview of Source Codes

First of all, we implemented the periodic sending of the message.

greet hello suite to a time or an event that can be managed according to
the Schedule.

[Link] Periodic Sending of the HELLO Message

void WsnSurvey::sendhello()
{
WsnHello *msg=new WsnHello("WsnHello",NETWORK_LAYER_PACKET); //Creation of the message
msg ->setSrcAdress(SELF_NETWORK_ADDRESS);
msg->setDestAdress(BROADCAST_NETWORK_ADDRESS);
msg->setHelloKind(HELLO); // Fill in the message parameters
trace() « "broadcast ["«msg->getName()« "] initialization from "«
msg->getSrcAdress() to msg->getDestAdress()
scheduleAt(simTime(), msg); //Schedule the periodic sending
toMacLayer(msg,BROADCAST_MAC_ADDRESS) ;// Send the message to the MAC layer
}

Once the HELLO Message is broadcasted, each node receiving this

Message must respond by sending a HELLO-Rep message for
confirm their presence as being a neighbor. And from this principle that the
the neighbors table is filled according to the following Code:
Implementation and Simulation 99

void WsnSurvey::updateTable(string s, string d, int r) {

int role=1 ;
NTable::iterator pos; pos = [Link](s);// Traverse the table to detect if the neighbor exists
if (pos == [Link]()) //Add a new Neighbor
{
Neighbor newEntry;
[Link]=d ;
[Link]=s ;
[Link]=r ;
nbvois=nbvois+1 ;
[Link](make_pair(s,newEntry)); //Insert the Neighbor
trace()"neighbors"<nbneighbors;

trace()«"nveau voisin" «[Link]«"et"«[Link] ;

}
else
Neighbor exists, update the role
Neighbor Entry=pos->second ;
[Link]=d ;
[Link]=r ;
Neighbor already exists
}
role = RoleAttribute(); // Role Assignment
WsnHello * msg=new WsnHello("WsnHello", NETWORK_LAYER_PACKET);
msg->setRole(role);
}

[Link] Attribution of Roles

Once the table is filled by a new neighbor, this node Com-

begins to assign roles according to the distribution conditions of the cap-
All of this can be summarized in this Code:
Implementation and Simulation 100

intWsnSurvey::RoleAttribute()
{
trace()"attribution";
intnbLiaison=0 ;intnbRep=0 ;intnbSimple=0 ;
for (NTable::iterator pos = [Link](); pos != [Link](); ++pos)
{
Neighbor aNeighbor = pos->second;
if([Link] == 3)
{
// Compter le nombre de representants
nbRep++;
if([Link] == 2) nbLiaison++ ;
if([Link] == 1) nbSimple++ ;
}
trace()«"nbre repr"«nbRep«"et nbre liai "«nbLiaison«"et simple"«nbSimple ;
Exact start of the algorithm
if(nbRep == 0) IchRole = 3 ;//link node
else
{
if(nbLiaison== 0) IchRole=2 ;//representative
else
{
if(nbRep > 1) IchRole = 1 ; //simple node
}
}
//fin
trace()«IAmRole; // Display the role of the node
return IchRole ;

[Link] Management of redundancy

Using the property of the Gabriel graph, the coverage of

the network can only be secured with the Representative and the nodes
active. Each group representative outlines the summarized process
in (Figure 4.8).
Once the roles have been assigned, the Representative node takes
the hand and starts the procedure to put the nodes to sleep
redundant.
Implementation and Simulation 101

FIG. 4.8 - Redundancy organigram of simple nodes.

The following algorithm is as follows:

voidWsnSurvey :: redundant()
{
if(IchRole==3) {// IchRole is a
//variable assigned to the role of the node
for(NTable::iterator pos=[Link](); pos != [Link](); ++pos)
{
Neighbor aNeighbor = pos->second;
string* des=pos->[Link] ;
if(([Link] == 1)&&(nbRep !=0)) { state = new cPacket("state", STATE) ;// Packet
//defines that the mode is SLEEP
toMacLayer(state, resolveNetworkAddress(des)) ;
if(([Link] == 1)&&(nbSimple !=0)) { state = new cPacket("state",STATE) ;
toMacLayer(state,resolveNetworkAddress(des)) ;
}
}
}}
Implementation and Simulation 102

[Link] Redistribution of Hello

After each period of time, the Representative node sends a

Hello message to rediscover your neighbors:
The variable START defines the time interval of the Replay.
voidWsnSurvey : :timerFiredCallback(int index)
{
if(index !=START) {
trace() "error" ;
return ;
}
if(index==START)
{
setTimer(START, interval);
sendhello();
}
}

4.4.3 The .ini file

This file allows the user to configure these model of si-

formulation for execution. The editor is aware of all
the NED declarations (simple modules or composite modules).
Implementation and Simulation 103

[General]
Always include the
main [Link] file
#============================================= include
../Parameters/[Link]
sim-time-limit = 100s Simulation Time
SN.field_x = 80 # meters
SN.field_y = 80 # meters
These tests include 3 nodes each,
coordinates will be specified manually
[Link] = 100 Node name
[Link] = "uniform"
Deployment Type # important wireless channel switch to allow mobility
#[Link] = true Mobility Management
#[Link] = 0
#[Link] = 0
Choose a radio and set the Tx power to a low value so
that node’s mobility has a better effect on connectivity
[Link][*].[Link] ="../Parameters/Radio/[Link]"
[Link][*].[Link] = "-5dBm"
#[Link][*].[Link] = "IDEAL" Radio Channel Type
[Link][*].[Link] = 0
These tests use big packets (2kb) to show interference clearly
This requires setting the max packet size in all communication layers
[Link][*].[Link] = "WsnSurvey" The Network Layer
[Link][*].[Link] = 2500
Throughput test application is used to send 2000-byte
packets to node 0 (which by default is the receiving
# node for this app). 5 packets per second will be sent
[Link][*].ApplicationName = "BridgeTest" The Application Layer
[Link][*].[Link] = "BypassMAC" The MAC layer
[Link][*].Application.packet_rate = 5
[Link][*].[Link] = 2000
application’s trace info for node 0 (receiving node)
# is turned on, to show some interesting patterns
Specify the information gathering at the layer levels
[Link][*].[Link] = true
[Link][*].[Link] = true
[Link][*].[Link] = true
[Link][*].[Link] = true
[Link][*].[Link] = true
[Link][*].[Link] = true
Implementation and Simulation 104

4.5 Results Obtained

4.5.1 Network Deployment

The disadvantage of Castalia is that it does not allow for a

view of the network during execution knowing that the researchers are
working on it to enrich the Simulator especially after its evaluation.
evaluation among the safest Simulators.
For this, these figures show the graph of the deployment of
nodes at the network level in the Initial state.

FIG. 4.9 – Deployment of the nodes.

Implementation and Simulation 105

FIG. 4.10 – Execution with the Deployment.

subsectionFile Trace After a correct compilation (without de-

error detection) of Castalia with the new parameters such as
previously mentioned, Two files are created:
– Trace File: The latter provides a general and detailed overview of
process followed by each node during the simulation.
– Summary File: This shows the state of the node by summarizing
for example the number of packets received, lost, and the consumption
energy automation.

[Link] Initialization of nodes

An overview of the file during initialization or deployment of

nodes at the network level.
Implementation and Simulation 106

0 [Link][0].MobilityManager initial location(x :y :z) is 43.9051 :47.4276 :0

0 [Link][0].[Link] Initialized RX mode to normal

0 [Link][0].[Link] Initialized TX power output to -5 dBm, consuming 26.7 mW

0 [Link][0].[Link] Default sleep level initialized to idle

0 [Link][0].[Link] completing transition to 0 (RX)

0 [Link][1].MobilityManager initial location(x :y :z) is 67.5413 :48.2211 :0

0 [Link][1].[Link] Initialized RX mode to normal

0 [Link][1].[Link] Initialized TX power output to -5 dBm, consuming 26.7 mW

0 [Link][1].[Link] Default sleep level initialized to idle

0 [Link][1].[Link] completing transition to 0 (RX)

0 [Link][2].MobilityManager initial location(x :y :z) is 43.5907 :67.7801 :0

0 [Link][2].[Link] Initialized RX mode to normal

0 [Link][2].[Link] Initialized TX power output to -5 dBm, consuming 26.7 mW

0 [Link][2].[Link] Default sleep level initialized to idle

0 [Link][2].[Link] completing transition to 0 (RX)

0 [Link][3].MobilityManager initial location(x :y :z) is 49.8851 :51.6715 :0

0 [Link][3].[Link] Initialized RX mode to normal

0 [Link][3].[Link] Initialized TX power output to -5 dBm, consuming 26.7 mW

0 [Link][3].[Link] Default sleep level initialized to idle

0 [Link][3].[Link] completing transition to 0 (RX)

0 [Link][4].MobilityManager initial location(x :y :z) is 35.007 :23.8028 :0 0.

.
.
.
.
.

[Link] Neighborhood recognition

During this period, all sensors switch to RX mode and

send a Hello message for neighborhood recognition including
Implementation and Simulation 107

we can see it in the trace file:

0.000937365 [Link][11].[Link] broadcast [WsnHello] in initialization from 11 to -1

0.000937365 [Link][11].[Link] message transmission to the Mac layer with dest -1 and source 11

0.000937365 [Link][11].[Link] Buffered [BypassRouting packet] from MAC layer

0.000937365 [Link][11].[Link] SET STATE to TX, delay=1e-05, power=22.2

0.000947365 [Link][11].[Link] completing transition to 1 (TX)

0.000947365 [Link][11].[Link] Sending Packet, Transmission will last 0.00583333 secs

0.000947365 [Link][10].[Link] START signal from node 11, received power -92.9274dBm

0.000947365 [Link][10].[Link] START signal from node 11, received power -92.9274dBm

0.000947365 [Link][7].[Link] START signal from node 11, received power -98.1536dBm

0.000947365 [Link][7].[Link] Failed packet (WC_SIGNAL_START) from node 11,

0.000947365 [Link][6].[Link] START signal from node 11, received power -91.9877dBm

0.000947365 [Link][5].[Link] START signal from node 11, received power -90.7127dBm

0.000947365 [Link][3].[Link] START signal from node 11, received power -90.3762dBm

0.000947365 [Link][2].[Link] START signal from node 11, received power -79.5319dBm

0.000947365 [Link][0].[Link] START signal from node 11, received power -91.1788dBm

0.002031581 [Link][13].[Link] broadcast [WsnHello] in initialization from 13 to -1

0.002031581 [Link][13].[Link] message passage to the Mac layer with dest -1 and source 13

0.002031581 [Link][13].[Link] Buffered [BypassRouting packet] from MAC layer

0.002031581 [Link][13].[Link] SET STATE to TX, delay=1e-05, power=22.2

0.002041581 [Link][13].[Link] completing transition to 1 (TX)

0.002041581 [Link][13].[Link] Sending Packet, Transmission will last 0.00583333 secs

0.002041581 [Link][14].[Link] START signal from node 13, received power -90.9519dBm

0.002041581 [Link][8].[Link] START signal from node 13, received power -95.2822dBm

.
.
.
.
Implementation and Simulation 108

[Link] Neighbourhood Table and Role Assignment

Once the nodes receive the HELLO message, each node

fills its neighborhood table and at that moment, each node takes
a role suitable for him according to his position within the network.
This process is executed at the beginning after the recognition of the neighbor.
swimming and the allocation of roles.
In case of a role change due to the death of a node, it is
the sensor Representative who becomes responsible for this change
and assigns a role to the other nodes after consulting its table of
neighborhood.

0.01620540 [Link][14].[Link] TX finished (no more pkts in the buffer) changing to RX

0.01620540 [Link][14].[Link] SET STATE to RX, delay=1e-05, power=22.2

0.01621540 [Link][14].[Link] completing transition to 0 (RX)

0.01621540 [Link][1].[Link] message reception with destination 1 and source 14

0.01621540 [Link][1].[Link] Delivering [WsnHello] to Network layer

0.01621540 [Link][1].[Link] Received [WsnHello] from MAC layer

0.01621540 [Link][1].[Link] start of HELLRep message sending with source 1 and dest 14

0.01621540 [Link][1].[Link] table filling with source 14 and dest 1 and role 1

0.01621540 [Link][1].[Link] new neighbor 1 and 14

0.01621540 [Link][1].[Link] number repr 0 and number link 0 and simple1

0.01621540 [Link][1].[Link] role is 3

0.01621540 [Link][1].[Link] before sending the role is 3

0.01621540 [Link][1].[Link] message passage to the Mac layer with destination -1 and source 1

0.01621540 [Link][1].[Link] Buffered [BypassRouting packet] from MAC layer

0.01621540 [Link][1].[Link] SET STATE to TX, delay=1e-05, power=22.2

0.01621540 [Link][1].[Link] Buffered [BypassRouting packet] from MAC layer

.
.
.
Implementation and Simulation 109

[Link] Transition to Sleep State

Here is a snippet of code that provides an overview of the state of a node.

when the latter receives the State Change Message to proceed
the Wake mode in Sleep mode for the purpose of 'Energy saving' like
previously mentioned.

0.129737196712 [Link][2].[Link] reception of the message with destination 2 and source 0

0.129737196712 [Link][2].[Link] Delivering [state] to Network layer

0.129737196712 [Link][2].[Link] Received [state] from MAC layer

0.129737196712 [Link][3].[Link] message reception with destination 3 and source 0

0.129737196712 [Link][3].[Link] Delivering [state] to Network layer

0.129737196712 [Link][3].[Link] Received [state] from MAC layer

0.129737196712 [Link][3].[Link] SET STATE to SLEEP, delay=5e-05, power=0.5

0.129737196712 [Link][2].[Link] message reception with destination 2 and source 11

0.129737196712 [Link][2].[Link] Delivering [state] to Network layer

0.129737196712 [Link][2].[Link] Received [state] from MAC layer

0.129737196712 [Link][3].[Link] reception of the message with destination 3 and source 11

0.129737196712 [Link][3].[Link] Delivering [state] to Network layer

0.129737196712 [Link][3].[Link] Received [status] from MAC layer

0.129787196712 [Link][3].[Link] completing transition to 2 (SLEEP)

4.5.2 Example of a Resume file

This file provides a summary of the collected information.

end of the simulation execution. For example, the energy consumed-
myself, the number of packets sent, received and even lost from each
node present in the network.
Implementation and Simulation 110

Castalia| module :[Link][0].ResourceManager

Castalia| simple output name :Consumed Energy

Castalia| 0.604138

Castalia| module :[Link][0].[Link]

Castalia| simple output name :RX pkt breakdown

Castalia| 1 Failed, below sensitivity

Castalia| 115 Failed, non RX state

Castalia| 17 Received with NO interference

Castalia| simple output name :TXed pkts

Castalia| 15 TX packets

Castalia| module :[Link][1].ResourceManager

Castalia| simple output name :Consumed Energy

Castalia| 2.82057

Castalia| module :[Link][1].[Link]

Castalia| simple output name :RX pkt breakdown

Castalia| 1 Failed with NO interference

Castalia| 10 Failed, below sensitivity

Castalia| 86 Failed, non RX state

Castalia| 20 Received with NO interference

Castalia| simple output name :TXed pkts

Castalia| 17 TX packets
Implementation and Simulation 111

4.5.3 Simulation Parameters

The sensor deployment is random in a square area.

the dimension varies according to the number of nodes. The range of
communication as well as the initial value of energy are fixed for
all the nodes.

The two tables below provide more details:

Transmission range 20 m
Initial Energy 18720 Joules (equivalent to 2 AA batteries)
Antennas Bidirectional
Duration of Simulation Varies according to the number of nodes

Application Bridge Layer Test

Network Layer WsnSurvey
Phy Layer + MAC ByPassMac
Radio CC1000

[Link] The Radio CC1000

Define the Transition Matrix. The latter specifies the deadlines.

to switch between the 3 main states of Radio (Rx, Tx,
Sleep

RX TX Sleep
RX - 0,01 0,2
TX 0.01 0.2
-
Sleep 0.05 0.05 -
Implementation and Simulation 112

4.6 Results

First of all, we will provide an overview of the behavior

statement of the nodes within the network and their organization, then
we will detail the results obtained graphically.

4.6.1 Organization of the Network

Following the absence of the Castalia simulator that is presented in the

visual part of the network behavior. We chose to tra-
value more on the data collected in the trace files for
to illustrate an idea about the organization of sensors within the
network to at least validate the algorithms used.
We simulated a network composed of 50 nodes, deployed
randomly. Following that, we studied the trace file in detail.
generated first by the following points:
The positions of each node;
The neighbors discovered;
Defined roles.

[Link] The positions of the nodes

The following table shows the positions of the extracted nodes at

extracted from the generated trace file.

Nodes 0 1 2 3 4 5 6
Positions43.9,47.4 67.5,48.2 43.5,67.7 49.8,51.6 35.0,23.8 4.53,77.0 30.6,38.2

.
.
Nodes 20 21 22 23 24 25 26
Positions21.1,14.9 58.9,36.4 45.4,10.8 25.9,49.4 48.9,17.7 30.9,75.4 54.5,35.9
Implementation and Simulation 113

.
.
Nodes 43 44 45 46 47 48 49
Positions75.9,29.4 65.6,1.08 49.8,67.0 7.68,77.7 70.2,37.4 78.1,4.45 36.0,59.1

[Link] The neighborhood

The table illustrates the neighbors of some essential nodes such as

what:
Representative Nodes: 22.47
Link Nodes: 5, 43

NeighboringNodes
22 4,7,8,9,12,17,20,21,24,26,27,30,38,40,43,44,47,48
47 1,2,3,7,11,13,14,15,18,21,24,26,27,29,37,41,43,45
43 1,3,8,9,14,15,17,18,21,22,24,26,27,29,30,35,37,41,47,
15 0,2,3,5,6,8,21,23,24,25,26,27,32,33,34,36,38,41,43,45,

FIG. 4.11 – Random Deployment.

Implementation and Simulation 114

FIG. 4.12 - Organization of Nodes.

According to the diagram, we can see the organization of the nodes within
of the network after the deployment and following the recognition of the neighbor
swim by exchanging the messages Hello and HelloRep, as well as with the exe-
execution of the algorithm 'AttribeRole' mentioned earlier for the struc-
saturation of the network in Multi-Agent systems.

4.6.2 Node Redundancy

Once the organization of the network stabilizes, the re- algorithm

The dance begins its execution.
As a result, the redundant nodes become inactive while remaining
a network coverage.
The following diagram is the succession of the 2 previous ones, from which we notice
that the redundant nodes have gone into Sleep mode.
Implementation and Simulation 115

FIG. 4.13 – Redundancy Management.

4.6.3 Network Reconfiguration

In this case, we caused a node failure.

Simple to see the network reconfiguration.
At a moment 'T' we put a node in 'Sleep' mode to ge-
generate a stop of this sensor due to 'energy depletion' for example.
Following this, we can see the network reconfiguration after the dispa-
ration of this active node and the awakening of another Simple node to co-
open the area that was previously covered by this malfunctioning sensor.

FIG. 4.14 – Reconfiguration following a failure.

Implementation and Simulation 116

4.6.4 Energy Consumed at the Network Level

Energy consumption result according to the adopted monitoring method

fueled by the distribution of a number of different nodes for a
same topology

surface of 100*100,
Energy of 5 mJoule,
Number of nodes: 20, 50 up to 100
.

FIG. 4.15 – Energy Consumed in Network.

It is noted that energy consumption doubles until the

triple by increasing the number of nodes and this is mainly due to
to the messages exchanged for network stability.
Control messages increase at the expense of increasing
sensors to the network for the proper organization of the system.
Implementation and Simulation 117

4.6.5 Energy Consumed With and Without Monitoring

According to the histograms recorded for the different number of

nodes in a network, we can conclude that the average energy consumed
My following the Surveillance method is low compared to the...
a little unattended and it is due to the inactive states that the
redundant nodes.
Case 1: 20 knots

FIG. 4.16 – Energy Consumed at the Level of the Two Methods

Case 2: 50 knots

FIG. 4.17 – Average Energy Consumed by 50 nodes

Implementation and Simulation 118

Case 3: 100 nodes

FIG. 4.18 – Average Energy Consumed per 100 knots

4.6.6 Remaining energy of a representing node

Based on the organization of the nodes, we can deduce that the node
"Representative" consumes more energy than all other nodes,
what becomes interesting to study instead of simulating the state of energy
Tie this knot.

FIG. 4.19 – The remaining energy at a node Representing

Implementation and Simulation 119

These two curves demonstrate that the remaining energy at a level of a

nœud Représentant différe au niveau des deux méthode. et plus pré-
Currently in the surveillance method, the energy lasts longer than
the normal method due to the sending of inter-cluster messages with
a smaller radius and a reduced number of nodes.

4.6.7 Average number of packets from the different nodes

This curve presents an average of the packets sent and received.

par différents nœuds selon leurs rôles où on résume que les nœuds
représentant ont une moyenne plus élevés puisque ils responsable
on the organization of the network.

FIG. 4.20 – Average packets sent and received by nodes.

4.6.8 Energy Consumed by the Different Nodes

The two results provide an insight into energy consumption.

give at the level of some nodes for the 2 configurations for a
comparison from which we can conclude that energy consumption
different and that in the case of 'With Surveillance' it is much more op-
I'll take the case "Without Supervision."
Implementation and Simulation 120

FIG. 4.21 – Energy Consumed With Monitoring.

FIG. 4.22 – Energy Consumed Without Supervision.

4.6.9 Influence of Node Mobility on the Network

We conducted another simulation case regarding the mo-

node visibility to see the influence of movement inside
of a network of a few sensors on the network.

Our test concerns the mobility of nodes 0, 5, and 10. The mobility is
done just now 100 for different distances for each node
with a speed of movement of 15. Thus we obtained the
Next result :
Implementation and Simulation 121

FIG. 4.23 – Influence of Mobility on the Network.

Through this test, we notice that the mobility of the nodes has a very
great influence on the network in the case of energy rate, hence
note that energy consumption in a network with mobi-
It's much higher than in a network without mobility.
mainly due to the reconfiguration of the network and the-
Send messages for the discovery of the new neighborhood until
the stability of the network.

FIG. 4.24 – Influence of Mobility on the Network (Average Energy).

Implementation and Simulation 122

4.6.10 Influence of Simulation Time

Simulation time plays an important role in behavior

sensor management, for this we played with the time of sleep
formula to assess the network's operational duration by-
regarding the energy rate.

FIG. 4.25 – Remaining Energy relative to Simulation Time.

FIG. 4.26 – Average Remaining Energy.

Implementation and Simulation 123

4.7 Conclusion

Our goal, through the simulations carried out, was to test the per-
performances of our "Network Surveillance" Approach from which we
we were able to show the advantage of the distributed method to ensure
Fault tolerance with redundancy management.

On the other hand, this experience has pushed us to know the new
veal Simulated OMNeT++ and its Castalia platform and has allowed us to
integrated our approach and enrich the Simulator with new
new protocols.

Following the obtained results, we have observed that the method

simulated is more effective in energy consumption within the re-
This enhances the lifespan of the network.

Following all these benefits achieved with this approach, the research
futures are more favored towards the distributed approach with the aspect
from redundancy within the network for Surveillance management.
General Conclusion

Energy consumption at the levels of the Networks of Cap-

tensions is a sensitive point, which motivated researchers to focus on
further clarify this issue in order to extend the duration
of the life of sensors as well as Networks.
As a result, several models for different types of topologies have
were proposed and processed by the simulation tool in order to acquire
to a better configuration in energy consumption.

Following this, our ideas were based on the proposal of a surplus

veillance auto-configurable au niveau des nœuds capteurs. Nous avons
simulate an organization system based on the principle of Agents
Representative Nodes
lasts re-organizes according to the network distribution from which we have
introduces the concept of Node Redundancy Surveillance and
in order to improve energy consumption within the network.
Graph theory has allowed us to implement this concept.
of redundancy in the network.

The simulation results are quite satisfactory following the use.

the new OMNeT++ Simulator with the 'Castalia' platform that
is more dedicated to Sensor Networks. The latter is based on the
principle of the modules. Based on our various tests conducted, we
we noticed the decrease in energy consumed at the nodes
which ensures the longevity of the network.
General Conclusion 125

Tirant benefits from the Multi-Agent approach, as a perspective of

our work, we plan to study the impact of the use of
other energy consumption metrics that take into account
global information about the network.
Bibliography

Azadeh Abdolrazaghi, "Unifying Wireless Sensor Network

Simulators, Master’s Degree Project Stockholm, Sweden, 2009.

Samira Allam, "Multi-agent approach to controlling flooding"

dation dans un réseau de capteurs", Mémoire de fin d’étude pour
the attainment of the state engineer diploma in computer science, School
Higher National School of Computer Science (ESI), Oued-Smar Algiers, 2009.

[ASS08] [Link],[Link],[Link], E. Cayirci,

Wireless sensor networks: a survey, Computer Networks, Vol. 38
pp. 393-422, 2002.
B:
[BAB06] M. Badet, W. Bonneau. "Setting up a testing platform"
and experimentation", Master's Project in Internet Technology
1st year, University of Pau and the Adour Countries. 2006.
Decentralized approach for the
sécurité d’un RCSF", Memoire de fin d’étude pour l’obtention d’un
Engineer diploma in computer science, University of BECHAR,
June 2010.
Rachid Bechar, "Redundancy Theory for Reconfiguration
tion of systems - Application to wireless sensor networks
Master's thesis, Doctoral School STIC, Abdelhamid University
Ibn Badis of Mostaganem. Department of Computer Science, 2009.
BIBLIOGRAPHY 127

Djawhara Benchaira, Ahlem Bencheikh, "Security of the dessi-

data mining in a wireless sensor network: case of the protocol
Tiny Diffusion, 2009.
J. Bredin, E. Demaine, [Link] Hajiaghayi, and D. Rus
Deploying Sensor Networks with Guaranteed Capacity and Fault
lerance", in the Proceedings of the 6th ACM International Sym-.
Symposium on Mobile Ad Hoc Networking and Computing (Mobi-
HOC'05), Urbana-Champaign, Illinois, 2005.
Athanassios Boulis. "Castalia: a simulator for Wireless Sen-
Boby Area Networks User’s Manual version 2.1
NICTA, July 2009.
[BOU10] A. Boulis, "Castalia, a simulator for wireless sensor networks
and body area networks, version 3.1", User’s manual, NICTA, De-
December 2010.
C:
Scott Carson, Joseph Macker, "Mobile ad hoc Networking
(MANET): Routing Protocol Performance Issues and Evaluation
Considerations", RFC 2051, 1999.
[CAS08] Claude Castelluccia, "The security of sensors and sensor networks"
teur", INRIA, 2008.
[CET05] Bilge Cetin,"SIMULATION ENVIRONMENT", Novembre
2005
Yacine Chellal, "Wireless Sensor Networks", Systems in-
Smart for transport, Version 1 SIT60, 2008.
C-Y. Chong and S.P. Kumar, "Sensor networks: Evolution,
opportunities, and challenges"Proceedings of the IEEE, Vol. 91, No.
8, pp. 1247-1256, 2003.
BIBLIOGRAPHY 128

Akyildiz, W. Su, E. Cayirci, Y. Sankarasubramaniam. 'A sur-

Survey on sensor networks", IEEE Communications Magazine, vol. 40,
no. 8, pp. 102-114, Georgia Institute of Technology, Atlanta, USA.
2004.
D:
N. Doufène, H. Hadjammar, End-of-study project for the
State Engineer Diploma: "Routing in Networks"
of sensors: Optimization of the Directed Diffusion protocol, Institute
National Computer Training (INI). 2006.
Witold Drytkiewicz, Steffen Sroka, Vlado Handziski, An-
dreas Köpke, Holger Karl, "A Mobility Framework for OMNeT++"
Telecommunication Networks Group Technical University Ber-
lin, 2003.
E:
[ELK] N.J El Karaki, A.E Kamel, "Routing technique in wireless sensor
network : a survey, dept. of electrical and computer engineering
Iowa State University.
F:
The ns Manual (formerly ns
Notes and Documentation), The VINT Project May 3, 2008
Contribution to the Diagnosis of Failures
Differentially Flat Systems
the attainment of the degree of Doctor of the University of Toulouse,
delivered by INSA Toulouse, 2007.
PM. Frank. "Fault diagnosis in dynamic systems using analyti-
cal and knowledge-based redundancy - a survey and some new results.
In Automatica, volume 26, pages 479-474, 1990.

G:
BIBLIOGRAPHY 129

Detection and diagnosis

for discrete event systems - towards the consideration of er-
modeling errors
applied sciences of Toulouse, CNRS Laboratory, 2007.H :
[HAM07] [Link], Master's thesis: "Agent-based platform
for the design and simulation assistance of wireless sensor networks without
National Institute of Computer Training (INI), 2007.
[HCB00]W. Heinzelman, A. Chandrakasan, H. Balakrishnan,
Energy-Efficient Communication Protocol for Wireless Micro Sensor
Networks, In proceedings of the Hawaii International Conference on Systems
Items Science, vol. 8, pp. 8020, January 2000.
Uwe Herzog, "Ad hoc Networks: New Life For An Old
Concept", [Link], 2005.
Ghaleb Hoblos, "Diagnosis and Surveillance of Complexes"
Specialty: Automation, Doctor of the university of sciences and
Lille technologies; Teacher: Researcher at L'ESIGELEC, 2008.
A short survey of wireless sensor
networks, Technical University Berlin, Telecommunication Net-
works Group, 2003.
K. Xu, H. Hassanein, G. Takahara and W. Wang, "Relay
node deployment strategies in heterogeneous wireless sensor networks
single-hop communication case", in the Proceedings of the IEEE
Global Telecommunications Conference (Globecom'05), St. Louis
MO, 2005.
I, J :
A Quick Overview of the OMNeT++4.0 IDE

K:
BIBLIOGRAPHY 130

Bouabdellah Kechar, "Issues of Consumption

Energy in wireless sensor networks
d'oran, 2007.
Tatiana Kempowsky, "Process monitoring based on mé-
Classification method: Design of a support tool for detection and
the diagnosis of failures
lysis and architecture of systems of the CNRS, In order to obtain
Awarding of the title of doctor from the national institute of applied sciences
questions of Toulouse. 2004

B. Khalifa. "Security in wireless sensor networks"

conference at the University of Bechar. 2006.

L:
Tayeb Lemlouma, Abdelmadjid Boudina, "Artificial intelligence
distributed science and multi-agent systems.
[LIL09] LIANG LI, "Implementing Transport Protocol PSFQ in Sensor
Network in Castalia", School of Computer Science, 2009
[LWI07] Marc Löbbers, Daniel Willkomm, "A Mobility Framework for
OMNeT++ User Manual Version 1.0a4, 2007.
M:
Sensor networks: localization,
verture et fusion de données.",THÈSE pour obtenir le grade de
Doctor from the University of Franche-Comté Specialty: Infor-
Thesis prepared at the Computer Science Laboratory of the University.
Franche-Comté site (LIFC) within the framework of the Doctoral School
Sciences for Engineering and Microtechniques (SPIM), 2008.
[MAT08] Pantélis Matsos,"Réseaux Sans Fil et de Mobiles RSFM", 2008.
BIBLIOGRAPHY 131

Didier Machine "Process Surveillance", Di-

Postgraduate Diploma in Control, Signals and
Communications Vanduvre-les-Nancy, February 2003.
R. Meraihi, "Management of service quality and topology control"
"logistics in ad hoc networks", Thesis presented for the award of the
Doctorate degree from the National School of Telecommunications
Communications of Paris, 2003.

[MOA08] Sofiane MOAD, Supervisor: Nizar Bouabdallah,"The

energy consumption in wireless sensor networks Master
Research 2 in Computer Science IFSIC-Rennes1, 2008.
[MSK05] C. Mallanda, A. Suri, V. Kunchakarra, S. S. Iyengar*, R. Kan-
nan* and A. Durresi"Simulating Wireless Sensor Networks with
OMNeT++, S. Sastry The University of Akron, Akron, Ohio.
[MSP09] Abdallah Makhoul, Rachid Saadi, Congduc Pham, "Sur-
video surveillance on wireless sensor networks: adaptive scheduling
tatif avec prise en compte de la criticité", LIUPPA, Université de Pau
and the University of the Adour (UPPA).

N,O :
OMNeT++, Discrete Event Simulation System Version 4.0
User Manual
[OMN10]" OMNeT++ Installation GuideVersion 4.1",2010P,Q :
Pujolle G., "The Networks Editions 2005", Eyrolles editions
2005.
R:
Patrick Ripoll, "Design of a fuzzy diagnostic system
"applied to the automobile engine", to obtain the grade of Doctor of
The University of Savoie (Specialty Electronics, Electrotechnics
and Automatic), 1999.
BIBLIOGRAPHY 132

[ROM07]Yasser Romdhane,"Evaluation of the performance of proto-

SMAC and directed diffusion in sensor networks
final report, Higher School of Communication of Tunis,
2007.
S:
Energy Efficient Protocol (EEP): a protocol for
Energy-efficient routing for wireless sensor networks, Thesis
of studies For the obtaining of the state engineering degree in
computer science; 2008/2009.
T,U,V :
Andreas TIMM-GIEL, Ken MURRAY, Markus BECKER
CiaranLYNCH, Carmelita GÖRG, Dirk PESCH,"Comparative Si-
Simulations of WSN, ICT-MobileSummit 2008.
W,X :
Ding X., FrankP.M., "Frequency domain approach and thre-
should selector for robust model-based fault detection and isolation", Proc.
of IFAC Symp. SAFEPROCESS 91, Baden Baden, Germany, vol. 1,
pp.307-312, 1991.
Y,Z:
Mohamed Younis, Kemal Akkaya, 'Strategies and Techniques for'
Node Placement in Wireless Sensor Networks: A Survey
Department of Computer Science and Electrical Engineering

Applying Image Processing Techniques

Processing Techniques To Simulate a Self-organized Sensor Network For
TrackingObjects,MasterThesisDefense,ComputerEngineering
2005.
BIBLIOGRAPHY 133

Webography
[Link] documentation site in-
computing, September 2005.
[Link] site of the R and D division of
France Telecom, September 2005.
[Link] site of the federal polytechnic school
Lausanne, September 2005.
[Link] information site on the
simulatorAvrora, November 2005.
[5] http ://[Link]/nsnam/ns/
Summary
Wireless sensor networks are attracting more attention from the community.
research. Thus, simulation is a commonly used approach to test and
to validate approaches, simulation environments must be able to support
offfdifferent models of wireless sensor networks.
Our thesis focuses more specifically on the case of monitoring networks of
sensors. The protocol tested is a protocol based on the reconfiguration of the nodes with
the principle of redundancy to maintain coverage of the area and ensure a long
life of the network.

The validation of the approach is considered following the use of the simulator.
OMNeT++ since the results obtained showed that the concept of the approach to
level of fault tolerance (with the principle of waking up sleeping nodes in
failure case or energy depletion of an active node) ensures proper functioning
thus a longevity of the network.

Keywords
["Wireless sensor network","Monitoring","Redundancy","Reconfiguration","Simulation"]
OMNET++; Castalia; Routing; Network Layer; Neighborhood.