Energy Efficiency Routing Protocols for Wireless Sensor Networks: Contribution to lifetime maximisation

mindasparagusΔίκτυα και Επικοινωνίες

14 Ιουλ 2012 (πριν από 5 χρόνια και 1 μήνα)

2.349 εμφανίσεις

UNIVERSITÉ MOHAMMED V

AGDAL

FACULTÉ DES SCIENCES

Rabat


Faculté des Sciences, 4 Avenue Ibn Battouta B.P. 1014 RP, Rabat

Maroc

Tel +212 (0
5
) 37 77 18 34/35/38, Fax: +212 (0
5
) 37 77 42 61, http://www.fsr.ac.ma


N° d’ordre

:
2490


THÈSE
DE DOCTORAT


Présentée
par

Ouadoudi

ZYTOUNE





Discipline

: Sciences de l’ingénieur

Spécialité

: Informatique et télécommunication



Energy E
ffi
ciency Routing P
rotocols for Wireless

Sensor Networks: Contribution to lifetime
maxim
isation.



Soutenue le

: 2
4

Avril
2010











Devant le jury

:


Président

:

M. Omar FASSI FEHRI, PES (FS, Rabat)


Examinateurs :

M. Driss ABOUTAJDINE
, P
ES
(FS, Rabat
)

M.
Ahmed
AIT OUAHMAN
, PES (ENS
A
,
Marrakech)

M. Mohamed HAMRI
, PES (FS, Rabat)

M. Moh
amed ESSAAIDI
, PES (
FS
,
Tétouan
)

M. Samir BENNANI
, P
H
(
EMI
, Rabat)

M. Youssef FAKHRI, PA (FS, Kenitra)










2
R´ESUM´E
Depuis ces dernieres annees,on constate un developpement tres rapide des reseaux de
communication sans ls.Ce developpement a conduit a un inter^et de plus en plus im-
portant vers les reseaux sans ls sans infrastructure specique.En parallele,le progres
de la miniaturisation des composants electroniques a fait emerger un type particulier
d'equipements dedie a la surveillance a distance.C'est ainsi que le marche des reseaux
et des applications sans l s'est considerablement developpe et la recherche dans le do-
maine des capteurs subit actuellement une revolution importante,ouvrant des perspec-
tives d'impacts signicatifs dans de nombreux domaines d'application (securite,sante,
environnement,securite alimentaire,fabrication,telecommunication,robotique,...).
Un reseau de capteurs sans ls se compose generalement d'un nombre important de dis-
positifs sans ls capable de realiser des mesures environnementales.Les exemples typiques
incluent la temperature,la lumiere,le bruit,l'humidite,etc.Ces mesures de capteurs sont
transmises a travers une transmission radio vers un noeud central appele Station de Base
qui constitue la passerelle vers une application utilisateur dediee qui prend des decisions
basees sur ces lectures de capteurs.Souvent,les capteurs necessitent un deploiement
dans des environnements hostiles,ou les nuds ainsi que les liens de communication sont
continuellement exposes a des contraintes importantes.Cet objectif est complique davan-
tage a cause de l'absence d'infrastructure de communication xe,en plus de limitations
materielles imposees par la taille miniaturisee des capteurs.L'energie est une donnee
essentielle dans le design d'un reseau de capteurs sans ls du fait que les noeuds sont
generalement alimentes par des batteries.Partant de la constatation que dans un nud,
la majeure quantite d'energie est consommee durant la transmission,l'eort de cette
these est de proposer des algorithmes de routage permettant de reduire la consommation
ii
d'energie tout en assurant la delivrance des donnees.Dans cette these,deux types de
protocoles de routages ont ete proposes:
La premiere contribution concerne le routage plat.Son principe est de ne laisser par-
ticiper au routage que les nuds dont l'energie residuelle est superieure l'energie residuelle
moyenne du reseau permettant ainsi,un epuisement equitable de l'energie des dierents
nuds du reseau.Trois contributions sont proposees dans la categorie de protocoles
hierarchiques.La premire propose un algorithme de formation de clusters bases sur
l'evaluation de la distance du nud la station de base en passant par tous les cluster-
heads.Ainsi,les nuds choisissent leurs clusters en se basent sur la distance de chaque
nud par rapport chaque cluster-head et l'eloignement de ces cluster-head de la station
de base.La deuxieme contribution exploite le fait qu'un nud distant de la station de
base consomme plus d'energie qu'un nud proche quand ces nuds sont elus cluster-
heads.Ceci dit que les nuds du reseau ne doivent pas avoir le m^eme cycle pour devenir
des cluster-heads.
Ainsi,on a propose d'avoir des cycles inversement lies la distance du nud de la station
de base.La troisieme contribution traite la reduction de l'energie consommee lors de la
formation des clusters.Ainsi,au lieu de changer les cluster-heads chaque periode de
transmission,notre idee consiste accorder a chaque nud elu comme cluster-head ce r^ole
plusieurs periodes consecutives avant de le ceder pour les autres nuds.
Mots cles:Resaux de capteurs sans l,Ecacite energetique,Duree de vie,Routage,
Transmission sans ls,Clustering.
ABSTRACT
Routing protocols in wireless sensor networks are a crucial challenge for which the goal
is maximising the system lifetime.Since the sensor nodes have limited capabilities,these
routing protocols should be simple,scalable,energy-ecient,and robust to deal with a
very large number of nodes.They should also be self-congurable to node failures and
changes of the network topology dynamically.The most proposed routing techniques
organize the network in clusters where the sensing area is divided into many sub-areas.
This thesis presents newalgorithms for routing in Wireless Sensor Networks that permit to
exploit the network energy to extend the network lifetime.In this tesis some contributions
in the area of routing were proposed.The rst one deals with at networks.The principle
behind this idea is to balance equitably the energy load among all the network nodes.
The transmission path is selected based on the minimum transmission energy and the
remaining residual energy of each node.The second contribution proposes an algorithm
the elect its cluser for each node considering the cluster-head energy load.The third work
gives a method to select the node as cluster-head based on its distance to the sink that
allows better balancing the cluster-head cost fairly among all the network nodes.Finally,
a clustering algorithmfor heterogeneous networks was propsed.This algorithmallows the
reduction of the cluster control energy,by making each selected cluster-head playing this
role for many consecutive rounds.An evaluation of the proposed techniques was done.
The simulation shows that the proposed algorithms give better performannces than the
compared ones.
AVANT-PROPOS
Ce travail a ete eectue au Laboratoire de Recherche en Informatique et Telecommunications
(LRIT) de la Faculte des Sciences de Rabat - Universite Mohammed V Agdal,Maroc.
Je tiens en tout premier lieu a exprimer ma profonde reconnaissance a Monsieur Driss
ABOUTAJDINE,professeur a la Faculte des Sciences de Rabat et responsable du LRIT,
qui a dirige cette these avec un inter^et constant et une grande competence et qui de plus
qui me fait l'honneur de participer cette commission d'examen.Je le remercie vivement
pour sa disponibilite,son soutien,ses precieux conseils,et les encouragements qui m'ont
permis de mener a bien ce travail.
Je suis tres honore par la presence de Monsieur Omar FASSI FEHRI,secretaire perpetuel
a l'academie Hassan II des Sciences et Techniques qui a accepte de presider le jury de ma
these,egalement tres honore par la presence de Monsieur Mohamed HAMRI professeur
a la Faculte des Sciences de Rabat et Monsieur Abdellah AIT OUAHMAN Directeur
de l'Ecole Nationale des Sciences Appliquees de Marrakech,qui ont accepte d'^etre les
rapporteurs de cette these.Qu'ils trouvent ici mes plus vifs remerciements pour l'eort
qu'ils ont fait pour lire mon manuscrit et l'inter^et qu'ils ont porte a mon travail.Ils
ont egalement contribue par leurs nombreuses remarques et suggestions a ameliorer la
qualite de ce memoire,et je leur en suis tres reconnaissant.Mes sinceres remerciements
vont egalement a Monsieur Mohamed ESSAAIDI,professeur a la facult des Sciences de
Tetouan,Monsieur Samir BENNANI,professeur a l'Ecole Mohammedia d'Ingenieurs a
Rabat et Monsieur Youssef FAKHRI,professeur assistant a la Faculte des Sciences de
Kenitra,pour leur participation au jury de cette these.
Mes remerciements vont aussi a tous mes collegues dans les laboratoires LRIT.En par-
ticulier,un grand merci a Monsieur Mohamed EL AROUSSI,Ahmed FAQIHI,Mounir
vi
AIT KAROUM,Rachid SAADANE et a Monsieur Abdelilah JILBAB,pour leurs aides
et conseils.
Je remercie chaleureusement mes parents mes freres,ma femme Saida et mes tres chers
amis pour leur soutien et pour leurs felicitations lors de mes reussites...
Enn merci a ceux que je n'ai pu citer mais qui ont toutes mes amities et mes remer-
ciements.
TABLE OF CONTENTS
1 Introduction.....................................1
1.1 Wireless Sensor Networks...........................2
1.1.1 Sensor networks characteristics....................3
1.1.2 Wireless Sensor networks applications.................9
1.2 Problem Statement...............................14
1.3 Performance Evaluation............................15
1.4 Thesis Outline..................................15
2 Wireless Sensor Networks Routing Protocols................17
2.1 Introduction...................................18
2.2 Routing protocols for wireless sensor networks................19
2.2.1 Flat routing protocols.........................20
2.2.2 Location-based routing protocols...................33
2.2.3 Hierarchial routing protocols......................36
2.2.4 Routing protocols based on protocol operation............48
2.3 Conclusion....................................53
3 Flat Networks Routing Algorithms Contribution..............55
3.1 Introduction...................................56
3.2 Lifetime Maximizing Algorithm for Wireless Sensor Networks.......58
3.3 A Decentralized Lifetime Maximizing Algorithm for Wireless Sensor Net-
works......................................60
3.4 Simulation and Results.............................61
viii
3.4.1 Simulation setting............................61
3.4.2 Simulation results............................61
3.5 Conclusion....................................66
4 Hierarchical Routing Algorithms Contributions..............69
4.1 Introduction...................................70
4.2 Energy Aware Cluster selecting algorithm for Routing in WSNs......71
4.2.1 Motivation................................74
4.2.2 Set-up phase..............................76
4.2.3 Steady-State Phase...........................77
4.2.4 Simulations and Results........................77
4.3 Cluster-head selection based node to sink remoteness............81
4.3.1 Motivation................................82
4.3.2 Clusters heads selection........................82
4.3.3 Simulation and Results.........................87
4.4 Reducing Cluster Control Energy technique for heterogeneous networks..92
4.4.1 Set-up phase..............................96
4.4.2 Steady-State Phase...........................97
4.4.3 Simulation and Results.........................97
4.5 Conclusion....................................103
5 Conclusion and future work...........................105
A Consumption Energy Model...........................109
B List of Publications................................113
References........................................115
LIST OF FIGURES
1.1 Typical node architecture...........................7
1.2 Examples of sensor nodes............................8
1.3 Typical network architecture..........................8
1.4 WSN applications taxonomy..........................10
1.5 Some WSN applications............................10
2.1 Routing protocols taxonomy..........................21
2.2 Flooding Implosion...............................22
2.3 Flooding overlapping..............................22
2.4 SPIN basic protocol operations........................24
2.5 Directed Diusion protocol operations....................26
2.6 the leader node gets all the readings,calculates the average and if it is
greater than a threshold sends it to the base station............31
2.7 Virtual grid and active nodes in GAF.....................34
2.8 Data ow in clustered network............................38
2.9 Level clustering in TEEN...........................41
3.1 Relative Network lifetime Evolution vs Threshold..............59
3.2 Relative Network lifetime Evolution......................62
3.3 Number of nodes still alive over transmission time..............63
3.4 Total Data Packets transmitted........................63
3.5 Total Network Remaining Energy.......................64
3.6 Transmission Energy per round........................65
x LIST OF FIGURES
3.7 Network remaining energy modelisation....................65
3.8 Network lifetime comparison..........................66
3.9 Network transmission energy per round....................67
3.10 Data packets transmitted to the Base station.................67
4.1 Cluster-based mechanism in LEACH........................74
4.2 LEACH cluster formation owchart........................74
4.3 The cycle of LEACH operations...........................75
4.4 Selection of cluster-head for transmission to the base station..............75
4.5 Network Lifetime until the rst node dies.......................78
4.6 Number of nodes still alive by transmission rounds..................79
4.7 Network lifetime with no compression.........................79
4.8 Network Remaining Energy.............................80
4.9 Energy per round..................................81
4.10 Average number of nodes per cluster.........................82
4.11 Cluster-heads comparison...........................85
4.12 Number of clusters vs transmission rounds..................88
4.13 Lifetime of a network with 100 nodes.The base station at (50m,175m)...88
4.14 Lifetime of a network with 100 nodes.The base station at (50m,200m)...89
4.15 Lifetime of a network with dierent node density...............89
4.16 Lifetime of a network with dierent base station remoteness fromthe network.91
4.17 Network transmission energy by round.....................91
4.18 Protocol owchart................................97
4.19 Total Network Energy for Clusters Forming in DEEC............98
4.20 Relative Network lifetime...........................100
4.21 Total Network Energy for Clusters Forming in CTRWSN..........100
4.22 Network lifetime................................101
4.23 Network Remaining Energy..........................101
4.24 Network lifetime until the rst node dies f = 0:2 and a varying from 0:5 to 5102
4.25 Network lifetime until the rst node dies a = 3 and f varying from 0:1 to 0:9102
A.1 Radio energy dissipation model............................109
ABR´EVIATIONS
ADC Analog to Digital Converter
ADV Advertisement message
APTEEN Adaptive periodic threshold-sensitive energy ecient
ASIC Application Specic Integrated Circuit
BFA Brute Force Algorithm
BS Base station
CADR Constrained Anisotropic Diusion Routing
CDMA Code Division Multiple Access
CH Cluster head
CPU Central Processing Unit
CSMA Carrier Sense Multiple Access
CSMA/CA Carrier Sense Multiple Access With Collision Avoidance
DAC Digital to Analog Converter
DC-DC Direct Current to Direct Current Energy Converter
DEEAC Distributive Energy Ecient Adaptive Clustering Protocol
DEEC Distributed Energy Ecient Clustering algorithm for heterogeneous wireless sensor networks
DD Directed Diusion
DirQ Directed Query Dissemination
DSR Dynamic Source Routing
EAD Energy-Aware Data-Centric Routing Algorithm
EAR Energy Aware Routing
EEHC Energy ecient heterogeneous clustered scheme for wireless sensor networks
xii LIST OF FIGURES
EEPROM Electrically Erasable Programmable Read-Only Memory
FDMA Frequency Division Multiple Access
FPGA Field Programmable Gate Array
GAF Geographic Adaptive Fidelity
GEAR Geographic and Energy Aware Routing
GPS Global Positioning Sustem
HEED Hybrid energy-ecient distributed clustering algorithm
IDSQ Information-driven sensor querying
IP Internet Protocol
LEACH Low energy adaptive clustering hierarchy
LELE Leader Election with Load balancing Energy in Wireless Sensor Network
MAC Meduim Access Control
MCU Microcontroller Unit
MMSPEED Multi-path and multi-speed routing protocol
PEGASIS Power-Ecient Gathering in Sensor Information Systems
QoS Quality of Service
RAM Random Access Memory
REQ Resquest message
SAR Sequential Assignment Routing
SEP Stable Election Protocol for clustered heterogeneous wireless sensor networks
SPEAR Sensor Protocol for Energy Aware Routing in Wireless Sensor Network
SPEED Stateless Protocol for End-to-End Delay
SPIN Sensor Protocols for Information via Negotiation
TDMA Time Division Multiple Access
TEEN Threshold-sensitive energy ecient
TTL time-to-live
WSN Wireless Sensor Network
Chapter
1
INTRODUCTION
This chapter gives the reader a background related to the eld of wireless sensor networks.
We start by dening what is a wireless sensor networks,after,we present the functions
of the various building blocks of a sensor node and provide some node features.Finally,
a wide scope of applications are described to illustrate how this technology is making
its presence felt in everyday life.Based on the requirements posed by sensor network
applications,we then highlight the main characteristics that should be present in any
protocol designed for sensor networks.
In recent years,with the advent of further miniaturization of electronics,many wire-
less networks have emerged.The development in these networks focuses in data rates
enhancement and power consumption reduction.Because of wireless Internet access be-
comes increasingly popular,higher data rates are solicited.So,either more bandwidth
must be allocated,or advanced signal processing has to be used to increase the spectral
eciency.Power consumption has always been a relevant issue in wireless networks.In
certain applications,this concern is important than the needs for high data rates.In
fact,some emerging wireless networks are designed for low data rate applications.Conse-
quently,they can use simple radios and aggressive power management to accomplish very
low power consumption.Among emerging wireless networks,the wireless sensor network
is yet another type where low power consumption is crucial than high data rates.
2 CHAPTER 1.INTRODUCTION
1.1 Wireless Sensor Networks
What is a Wireless Sensor Network?A wireless sensor network is a collection of sensor
nodes organized into a cooperative network.A sensor is a small device which observes
the environment of physical parameters such as temperature,pressure,relative humidity,
sound,vibration,motion or polluants,at dierent locations [Romer,2004,Haenselmann,
2006].And,Wireless Sensor Networks (WSN) are highly distributed networks of wireless
sensor nodes,deployed in large numbers to monitor the environment or system.Compared
to other categories of wireless networks,WSN possess two principal characteristics:multi-
hop transmission and constrained energy sources.First,since sensor nodes have limited
transmission ranges and organize themselves in an ad hoc manner,which means that
two wireless sensor nodes that can not reach each other directly transmit on other sensor
nodes to relay data between them.In general,data packets from the source node have to
traverse multiple hops before they reach the destination.Second,since sensors are usually
small and inexpensive,they are usually,battery powered and are often deployed in a rough
or hostile physical environment.Therefore,changing the batteries is a so dicult task,
as some networks may consists of hundreds to thousands of physically distributed nodes,
and in many case is even,impossible to node access to do this replacement.Consequently,
these nodes have constrained energy sources,and any protocols to be deployed in sensor
networks need to be aware of energy usage.
These two features have important implications to the fundamental performance limits of
WSN.With respect to the performance of WSN,the data transmission capacity and the
lifetime of the sensor networks are critical and in uential toward the design of optimal
deployment planning and data gathering techniques used such networks.
As cited above,the WSN can used multi-hop to gather data toward the base station.
That involves similitude between WSNs and mobile ad hoc networks (MANETs).In fact,
these networks are similar in some ways.However,the applications and technical require-
ments for the two systems are signicantly dierent in several respects [I.F.Akyildiz,
2002,C.S.Raghavendra,2004,McDonald and Znati,2000,G.H.Lynn,2003,B.Krish-
namachari,2002,K.Romer,2002,R.Shorey,2006]:
1.1.WIRELESS SENSOR NETWORKS 3
• In WSN,data generated by each node,are to travel from multiple sources to a
data recipient or sink rather than communication between a pair of nodes.This
implies that in WSN,the communication is multi-point to point and in MANET
this communication is point to point.
• In most applications,the sensor nodes themselves are static (even though the sensed
phenomena may be);which implies that the dynamics in the two types of networks
are not similar.However,in recent development,sensor nodes are increasingly
allowed to move and change their location to monitor mobile events,which results
in unpredictable and frequent topological changes.
• The data collected by multiple nodes are based on common phenomena.So,there
existes a potentially degree of redundancy in the data being transmitted by the
various sources in WSNs.Also,there is potentially some dependency on trac
event generation in WSNs,such that some typical random-access protocol models
may be inadequate at the queueing-analysis level.In MANETs is not generally the
case.
• A critical resource constraint in WSNs is energy;this is not generally the case in
MANETs,because the MANETs devices that handled by users can be replaced or
recharged relatively frequently.The scale of WSNs and the necessity for unattended
operation for periods of weeks or months implies that energy resources have to be
controlled very judiciously.
• A WSN consists of hundreds or thousands of devices using sensors to monitor condi-
tions at dierent locations.These devices collaborate among themselves to achieve
a sensing network.So,the number of nodes in WSN can be very important than
the nodes in a MANET.
1.1.1 Sensor networks characteristics
Figure 1.1 shows the basic diagram of wireless sensor node.These subsystems are con-
trolled by the operating system,using drivers,protocols and algorithms.In a sensor node
4 CHAPTER 1.INTRODUCTION
there are four essential parts [V.Raghunathan,2002]:processing subsystem;sensing
subsystem;transceiver subsystem and,power subsystem.
• The processing system usually constitutes of a controller and a memory unit.The
controller unit can either be a micro-controller unit (MCU),a eld programmable
gate array (FPGA) or an application specic integrated circuit (ASIC).The MCU
together with its on-board RAM,Flash and/or EEPROM provide the sensor node
the intelligence.It is responsible for control of the sensors,and execution of com-
munication protocols and signal processing algorithms on the gathered sensor data.
The use of software which is easily altered or replaced depending on the circum-
stances makes the MCU a very exible controller.The MCU also has the ability
to assign to itself a number of operating states in which parts of the MCU.In gen-
eral,four main processor states can be identied in a MCU:o,sleep,idle,and
active.In sleep mode,the MCU and most internal peripherals are turned o,and
can only be activated by an external event (interrupt).In idle mode,the MCU is
still inactive,but other peripherals are active,for example,the internal clock or
timer.In the active state,multiple substates may be dened based on clock speeds
and voltages.Within the active states,the CPU and all peripherals are active.
Moreover,MCU can use dynamic voltage scaling to alter the operating frequency
to t the application needs.However the exibility that the MCU provides comes
at a cost [C.Talarico,2005,Nikolaidis and Laopoulos,2002],with typical trade-
os towards energy-eciency and performance loses.The use of an ASIC in which
everything is constructed in haedware can provide a higher energy-eciency and
improved performance with a loss towards the exibility that the MCU provide.
• The sensor subsystemconstitutes of a number of sensors that collects data for further
processing by the micro-controller.It is the component that translates physical
phenomena to electrical signals.There exists a variety of sensors that measure
environmental parameters such as temperature,light intensity,sound,magnetic
elds,image,etc.The sensors can be classied as either analog or digital devices
and it depends on how its output is produced.An analog sensor must be connected
1.1.WIRELESS SENSOR NETWORKS 5
to ADC before the data can be processed,which is not the case for a digital sensor.
• The communication subsystems responsibility is to provide the sensor node with
communication capabilities,in such a way that the sensor nodes are able to com-
municate with other sensor nodes in its proximity.The communication medium
for the sensor node varies and depends on the actual application.In some cases
wired communication can be applied,but its applications are limited.A much
more interesting technique to use in this subsystem is a wireless communication
component.The wireless communication medium used for sensor networks includes
radio frequency,optical and acoustic (e.g.ultrasound) techniques [M.A.M.Vieira,
2003].The most common choice,when constructing a sensor node today,is to use
a radio frequency (RF) transceiver.This is due to the fact that it ts the require-
ments of most wireless sensor network applications.It consists of a short range
radio which usually has a single channel,a low data rate and operates at unlicensed
bands of 868870MHz (in Europe),902928MHz (in Americas) or near 2:4GHz
(global).There are several factors that aect the power consumption characteris-
tics of a radio,including the type of modulation scheme used,data rate,transmit
power (determined by the transmission distance) and the operational duty cycle.
For the energy consumption,four dierent states need to be distinguished:Trans-
mit,Receive,Idle and Standby modes.An important observation in the case of
most radios is that,operating in Idle mode results in signicantly high power con-
sumption,almost equal to the power consumed in the Receive mode [van Hoesel,
2007].Thus,it is important to completely shut down the radio rather than set it in
the Id le mode when it is not transmitting or receiving due to the high power con-
sumed.Another in uencing factor is that,as the radio's operating mode changes,
the transient activity in the radio electronics causes a signicant amount of power
dissipation.For example,when the radio switches from standby mode to Transmit
mode to send a packet,a signicant amount of power is consumed for starting up
the transmitter itself [A.Wang,2001].Since the radio transceiver must be in sleep
mode to achieve energy eeciency,when its not used,a wake-up mechanism will be
integrated to switch ON the radio when the node has to receive.The idea,with a
6 CHAPTER 1.INTRODUCTION
wake-up radio[Gu and Stankovic.,2005,Jr.da Silva,2001,Miller and Vaidya.,2004,
J.M.Rabaey,2000,L.Zhong,2001],is to have a specialised circuit that detects
when a packet is inbound and noties the appropriate component which in turn
schedules the activation of the main transceiver so it is able to receive the incoming
data.The goal with such a wake-up receiver is to have very low power consumption,
preferably less than 1W,and to be able to distinguish the destination address,so
that the main receiver is only activated for packets that are addressed to this sensor
node.Even without the use of address recognition this approach lowers the power
consumed by the transceiver enormously.So far,constructing a wake-up radio that
are able to distinguish the destination address has not been achieved.
• The power-supply subsystem provides power to the other components of the sensor
node and there exist a number of dierent methods in which this can be accom-
plished.These methods can be classied into the following categorises[A.Kansal,
2003]:power distribution;energy reservoirs;and power scavenging.Energy reser-
voirs:In this case,the energy is stored in some type of container and as energy
is consumed the available energy is decreased.The most commonly used of these
techniques are the battery,which is the usual way of powering a sensor node today.
The authors of [Holger Karl,2005] gave the energy density for some batteries,these
values are illustrated in Table 1.1.
Battery type
Energy density
Lithium Battery
2880J=cm
3
Rechargeable Lithium Battery
1080J=cm
3
Table 1.1:Battery energy density.
Power scavenging devices can be used to restore energy into rechargeable batteries
(See table 1.2 taken from [Holger Karl,2005]).However,currently these techniques
are an issue of research in order to t in sensor nodes.
Thus,the most commonly used technique is the battery,which is the usual way of
powering a sensor node today.DC-DC converter and voltage regulator are circuits
1.1.WIRELESS SENSOR NETWORKS 7
Energy source
Power density
Solar (outdoor,direct sunlight)
15mW=cm
2
Vibrations
0:01 0:1mW=cm
2
Acoustic noise
10 6mW=cm
2
at 75dB
Table 1.2:Sample Scavenging Power Density.
that can provide a constant voltage to the sensor node.
Since,battery technologies advance much more slowly than semiconductor technolo-
gies For example,the Li-ion battery energy density increases only 50% from 1995 to
2005 (280 watthours per liter in 1995 to 580 Whr/l in 2005) Conner [2005].While
in the same period of time,the number of transistors of Intel processors doubles
every 24 months [Hiremane].These facts make the need to develop highly ecient
communication and data management protocols for WSNs.
Also,additional components may be included depending on the specic application.As
supplementary components we can cite a location nding system to determine their po-
sition,a mobilizer to change their location or conguration (e.g.,antenna's orientation),
and so on.However,as the latter components are optional,and only occasionally used,
rarely they are considered in the literature studies.
Figure 1.1:Typical node architecture
In wireless sensor node,the much lot of energy is consumed in the radio communication.
A comparison of computation and communication costs has shown that transmitting
one bit over a distance of 100 m consumes approximately the same energy as executing
8 CHAPTER 1.INTRODUCTION
3000 instructions [G.Pottie,2000].Then,to reduce energy consumption the number of
communications should be minimized,even at the cost of increasing data processing.The
gure 1.2 gives examples of WSN nodes pictures.As depicted,the size of the nodes is
very small.
Figure 1.2:Examples of sensor nodes
Figure 1.3:Typical network architecture
To give reader an overview of sensor node hardware,Table 1.3 gives some marcket nodes
features:
A wireless sensor network is a collection of nodes organized into a cooperative network.
These sensor nodes are able to communicate with other neighbor nodes using wireless com-
munication technique and to collect data from a phenomenon or a set of phenomenons.
Because of phenomenon is the entity of interest to the end-user,the collected data are
1.1.WIRELESS SENSOR NETWORKS 9
Sensor node platform
Microcontroller
Transceiver
CPU
Clock freq (MHz)
RAM (kB)
Program memory (kB)
Type
Freq (MHz)
Max data rate (kbps)
weC[con,a]
Atmel AT90S8535
4
0.5
8
RFM TR1000
916.5
10
Rene mote [con,a]
Atmel AT90S8535
4
0.5
8
RFM TR1000
916.5
10
Rene2 mote[con,a]
Atmel ATmega163
4
1
16
RFM TR1000
916.5
10
MIT AMPS [R.Min,2000]
Intel StrongARM SA1100
206
16384
512
National Semiconductor LMX3162
2400
1024
Crossbow MICA [con,b]
Atmel ATmega128L
4
4
128
RFM TR1000
433,915
40
Crossbow MICA2DOT [con,b]
Atmel ATmega128L
4
4
128
Chipcon CC1000
315,433,915
38.4
Crossbow MICA2 [con,b]
Atmel ATmega128L
7.37
4
128
Chipcon CC1000
315,433,915
38.4
Crossbow MICAz [con,b]
Atmel ATmega128L
4
4
128
Chipcon CC1000
2400
250
EYES (Nedap) [L.F.W.van Hoesel,2003]
TI MP430F149
2
8
60
RFM TR1001
868.35
57.6
EYES (Innion) [V.Handziski,2004]
TI MP430F149
2
8
60
Innion TDA 5250
868-870
64
BTnode rev3 [con,c]
Atmel ATmega128L
7.37
244
128
Chipcon CC1000
433,915
38.4
Moteiv Tmote Sky [con,d]
TI MP430F149
8
10
48
Chipcon CC2420
2400
250
Ambient  Node [con,e]
TI MP430
4.6
10
48
Nordic nRF9E5
868,915
50
Ambient SmartTag [con,e]
8051
16
-
4
Nordic nRF9E5
868,915
50
Table 1.3:Nodes characteristics.
routed using other sensor nodes as routers to a sink node.The sink (or base station) is
a collector that sends end-user interests to the sensor network and collects data related
to these end-user interests from the network.It is normally a resourceful node having
unconstrained computational capabilities and energy supply.There can be single or mul-
tiple base stations in a network.Practically,the use of multiple base stations decreases
network delay and performs better using robust data gathering.Base station in a network
can also be stationary or dynamic.The dynamic base stations can in uence the routing
protocols greatly because of its changing position which will be not clear to all the nodes
in a network.The data of interest for a particular end-user is either available directly
to the end-user or it must be forwarded through a network before it is available to the
end-user.This is illustrated in Figure 1.3.
Depending on the sensor network application,the data delivery model to the sink can
be continuous,event-driven,query-driven and hybrid [R.Min,2001].In the continuous
delivery model,each sensor sends data periodically.In event-driven and query- driven
models,the transmission of data is triggered when an event occurs or a query is generated
by the sink.Some networks apply a hybrid model using a combination of continuous,
event- driven and query-driven data delivery.The routing protocol is highly in uenced by
the data delivery model,especially with regard to the minimization of energy consumption
and route stability.
10 CHAPTER 1.INTRODUCTION
Figure 1.4:WSN applications taxonomy
Figure 1.5:Some WSN applications
1.1.WIRELESS SENSOR NETWORKS 11
1.1.2 Wireless Sensor networks applications
Actually,the WSNs are low-cost,easy deployment and,self-conguring.This make them
desirable for various application classes as compare to other networks [E.Biagioni,2002,
2003,A.Cerpa,2001,A.Mainwaring,2002,D.C.Steere,2000,H.Wang,2003,H.Wang,
I.F.Akyidiz,2002,Salem and Mohamed,2006].As main application classes of WSN,
we can cite data gathering,event detection,object tracking and sink-initiated querying
(see Figure 1.5 redrawn from [K.Sohraby,2007] and Figure 1.4 redrawn from [J.Yick,
2008]).In the following,we give a brief list of applications areas where WSNs are or can
be deployed:
• Environmental monitoring:Remote sensing has usually used to perform large-scale
environmental monitoring.Remote sensing techniques have been traditionally con-
signed to the implementation of airborne or space-borne congurations,e.g.using
satellites.Although,satellites do not allow continuous monitoring of a particular
location.Since the quality of images taken by satellites are weather conditions
dependent,WSNs can be used to complement remote sensing data from satellites
as they can be used to observe environments at high spatio-temporal resolutions
[Thomas C.Henderson,2004].
• Precision agriculture:WSNs may be employed to facilitate the tasks of the farmer by
continuously monitoring parameters such as soil temperature,moisture and salin-
ity(e.g.Irrigation Management:Soil Moisture to monitor water delivery to spe-
cic Irrigation blocks,soil moisture monitoring for Stress irrigation to improve crop
quality,remote Irrigation Control (valve actuation,pump control),monitoring Irri-
gation line ow/pressure water delivery to correct destination).This would allow
the farmer to act immediately to conditions that are unfavorable to the growth of
crops and thus ensure a higher gain.WSN applications in agriculture are not just
limited to growing crops.Also,they are used to monitor animals.Currently sows
are tagged to help farmers that need to detect when a sow enters its heat period.
Because of it is known that a correlation exists between the movement of sows and
its heat period,using a sensor nodes with built-in accelerometers allow farmers to
12 CHAPTER 1.INTRODUCTION
track the sows remotely and would also help identify when a sow is in heat.
• Industrial Automation:In addition to being expensive,using wires can be constrain-
ing,especially when system moving parts are involved.Then the use of wireless
sensors allows a rapid installation to sense equipment and allows observe locations
that would not be accessible if cables were attached.Erlier,the use of wired sen-
sors was too dicult to be implemented in a production line environment.The use
of wireless sensors in these applications is enabling,allowing a measurement to be
made that was not previously practical.Other applications include energy control
systems,security,wind turbine health monitoring.
• Logistics and transport:With a rapid increase in the number of vehicles on the road
worldwide,intelligent road management systems are seen as the key to managing
transport infrastructure eciently.Some researchers describe a WSN that uses
magnetometers to detect the presence of vehicles,speed and also detect the vehicle
type.The information provided by the trac monitoring system can be used to
optimise trac light settings at urban intersections.Road users can also use this
information to better arrange their activities and adapt their routes.
• Medicine and health care:The WSNcould provide interfaces for disabled,integrated
patient monitoring.It can observe and detect aged people's behavior,e.g.,when
a patient has fallen.These small sensor nodes allow patients a greater freedom
of movement and allow doctors to identify predened symptoms earlier on[Young
Han Nam,1998].The small installed sensor can also enable tracking and monitoring
of doctors and patients inside a hospital.Each patient has small and lightweight
sensor nodes attached to them,which may be detecting the heart rate and blood
pressure.Doctors may also carry a sensor node,which allows other doctors to
locate them within the hospital.The work presented in [Georey G.Messier,2007]
points the use of wireless nodes for patient surveillances and gives a trac models
for the data packets genetated by WSN node monitoring body temperature and
electrocardiogram (ECG) data.
• Military command and control:The rapid deployment,self-organization and fault
1.1.WIRELESS SENSOR NETWORKS 13
tolerance characteristics of sensor networks make them a very promising sensing
technique for military applications.Military sensor networks could be used to detect
and gain as much information as possible about enemy movements,explosions,and
other phenomena of interest,such as battleeld surveillance,nuclear,biological and
chemical attack detection and reconnaissance.As an example,sensors help to locate
the source of incoming small arms re so that counterattacks can be launched against
snipers quickly and precisely.
• Maintenance:Sensors embedded into machines and structures enable condition-
based maintenance of these assets [A.Tiwari,2004].Typically,structures or ma-
chines are inspected at constant time intervals,and components may be repaired
or replaced based on their time duration in service,rather than on their working
conditions.This method is expensive if the components are in good working or-
der,and in some cases,scheduled maintenance will not protect the asset if it was
damaged in between the inspection intervals.Wireless sensing will allow assets to
be inspected when the sensors indicate that there may be a problem,reducing the
cost of maintenance and preventing catastrophic failure in the event that damage is
detected.Additionally,the use of wireless reduces the initial deployment costs,as
the cost of installing long cable runs is often prohibitive.
• Disaster prediction and management:The ability to predict upcoming disasters
could have a major impact on saving lives.Such predictions could help with the
continuous monitoring capability of WSNs.For example,sensor nodes equipped
with seismoacoustic sensors can be deployed to monitor the behaviour of active
volcanoes.The data collected from such networks could be used to predict the
eruption moments.WSNs are also being used to complement deep-water tsunami
detection mark to help improve the detection of tsunamis.In addition to predicting
disasters,WSNs can also be used in the management of disasters after they occur.
Some researchers intend to embed sensor nodes into the structures of buildings.
These nodes stay dormant until a breakdown is detected.They then collaborate
to help create a view of the collapsed interior identifying the structurally strong
14 CHAPTER 1.INTRODUCTION
support walls and cavities.The WSN is also able to locate survivors using heart
pulsation sensors.The collected information is then propagated to the rescue team
using energy-ecient routing algorithms.
• Home applications:The tiny sensor nodes can be embedded into furniture and
appliances,such as vacuum cleaners,microwave ovens and refrigerators.They are
able to communicate with each other and the roomserver to learn about the services
they oer.These room servers and sensor nodes can be integrated with existing
embedded devices to become self-organizing,self-regulated and adaptive systems to
form a smart environment.
1.2 Problem Statement
Since many of the proposed WSNapplication systems involve large networks,it is essential
to provide routing infrastructures that concurrently oer small routing state,small rout-
ing stretch,and robustness [R.Fonseca,2005,Y.-J.Kim,2005,Y.Mao,2007].Small state
is critical for scalability and eciency.With only a few kilobytes of memory in typical
sensor node,reducing the routing state will enable supporting large networks.In addi-
tion,with smaller routing state,a lower maintenance trac is implied,as these two are
usually correlated.Since WSNs knowledge topology and connectivity changes due to node
failures and environmental impact,robustness entails handling such changes eciently.
More specically,to minimize resource consumption and interruption of higher-level sys-
tem components,the routing infrastructure must handle the changes in the network with
minimal trac and latency.The critical importance of any wireless sensing node is to min-
imize its consumed power.Therfore,all network stack protocols must be energy ecient
especially the communications ones,because it is the data transmission that consumes the
most lot of energy [Townsend and Arms].Because of the node radio subsystem requires
the considerable amount of power,it is suitable to send data over the radio channel only
when required.The best way,in terme of energy consumption,must be selected for data
travel.Additionally,it is important to minimize the power consumed by the sensor itself.
Therefore,the hardware should be designed to allow the processing subsystem to e-
1.3.PERFORMANCE EVALUATION 15
ciently control power to the radio,sensor,and sensor signal conditioner.With Wireless
sensor networks constraint,the purpose of this research eort is to design,implement,
and test a new wireless sensor network (WSN) routing protocol which coordinates sensor
node transmissions to extend network service durations and conserve energy in memory-
and power- constrained devices.In this work,two kinds of problems are addressed,the
path selection in at networks to balance the energy load between the network nodes and
the clustering algorithms in hierchical networks to extend the network lifetime.
Through this thesis,we propose some contributions for routing algorithms in wireless
sensor networks.As given above,the metric used in this work,is the network lifetime.
Therefore,the performed algorithm is that giving the extended lifetime.
1.3 Performance Evaluation
The primary metric for evaluating the performance of a sensor network is lifetime,which
means the network service duration from deployment until the network energy nodes ex-
haustion,even in some applications,the network lifetime is dened until the rst node
run out its residual energy.The principal limiting factor for the WSN lifetime the energy
supply.Therefore,each node should be designed to manage its local energy as a means to
enhance total network lifetime.In many applications it is not the average node lifetime
that is critical,but rather the minimum node lifetime.For wireless security application,
every node must last for multiple years.And a single node failure would make a vulner-
able area in the security systems.One tell that nodes would be energy scavenged.But,
this energy is low,and yet,the nodes would be enegy constrained.Indeed,the enegy
consumed for transmitting a data packet should be reduced.Obviously,the technique
that requires the minimal transmission energy will extend the network lifetime.So,the
transmission energy may be used to assess the routing protocols.Moreover,any proposed
communication technique should be decentralized to deal with network auto-conguration
and to allow scalability.So,performance evaluation takes in account the technique be-
havior;the best technique is that completely decentralized.Finally,the communication
protocols must be adaptable to the network topology,in order to facilitate the nodes
16 CHAPTER 1.INTRODUCTION
deployement,which means that a communication protocol that implies a specic nodes
deployement is expensive.
1.4 Thesis Outline
The thesis is organized as follows:
Chapter 2:LITERATURE REVIEW
This chapter reviews the state of the art of WSN routing algorithms in the literature and
shows performance evaluation technics.
Chapter 3:Flat network routing proposition
In this chapter we propose an algorithm for routing in WSN that permits to extend
network lifetime.This algorithm considers the network as at,where all network nodes
have the same role.
Chapter 4:Hierarchical network routing propositions
Chapter 4 presents some contributions for routing in WSN based on network clustering.
Chapter 5:Conclusions and future works
Chapter 5 concludes the thesis by presenting conclusion and directions for future research.
Chapter
2
WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
Sommaire
2.1 Introduction.............................18
2.2 Routing protocols for wireless sensor networks.........19
2.2.1 Flat routing protocols........................20
2.2.2 Location-based routing protocols..................33
2.2.3 Hierarchial routing protocols....................36
2.2.4 Routing protocols based on protocol operation..........48
2.3 Conclusion..............................53
18 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
2.1 Introduction
A wireless sensor network is a sensor network that uses wireless radio as its physical
layer.This network enables users to remotely access information and services electron-
ically.They are extremely versatile and can be deployed to support a wide variety of
applications in many dierent situations,whether they are composed of stationary or
mobile sensor nodes.These sensors deployment depends on the nature of the application.
For example,in environmental monitoring and surveillance applications,sensor nodes are
typically deployed in an ad hoc fashion so as to cover the specic area to be monitored.
In health care-related applications,smart wearable wireless devices and biologically com-
patible sensors can be attached to or implanted strategically within the human body to
monitor vital signs of the patient under surveillance.Once deployed,sensor nodes self-
organize into an autonomous wireless ad hoc network,which requires very little or no
maintenance.Sensor nodes then collaborate to carry out the tasks of the application for
which they are deployed.Each sensor has wireless communication capability and su-
cient intelligence for signal processing and networking of data.However,sensor nodes are
constrained in energy supply and bandwidth.Such constraints,combined with a typical
deployment of a large number of sensor nodes have posed many challenges to the design
and management of sensor networks.These challenges necessitate energy-awareness at all
layers of the networking protocol stack.And especially at the network layer,the main aim
is to nd methods for energy ecient route setup and reliable relaying of data from the
sensor nodes to the sink so that the lifetime of the network is maximized.The character-
istics of the environment within which sensor nodes typically operate,coupled with severe
resource and energy limitation,make the routing problem very challenging.The objec-
tive of this chapter is to survey the state-of-the-art routing protocols for WSNs.Thus,a
taxonomy of the basic routing strategies used to strike a balance between responsiveness
and energy eciency,is provided.The principle of each routing class is given,and some
related works are exposed.
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 19
2.2 Routing protocols for wireless sensor networks
Although,wireless sensor networks and ad hoc networks have some similitudes,WSNs
have theirs own unique characteristics which create new challenges for the design of rout-
ing protocols for these networks.Since,sensors have very limited transmission power,
computational capacities and most of all,stored energy.The operating and networking
protocol must be much simpler as compared to other ad hoc networks.This simplicity may
also break with conventional layering rules for networking protocols,since abstractions
typically cost time and space.Also,due to the large diversity of application scenarios
for WSNs,it is impossible that there will be a one-thing-ts-all solution for these very
dierent possibilities.So,the application requirements involve the specic design of a
sensor network routing protocol.In fact,the data trac in WSNs has signicant redun-
dancy since data is probably collected by many sensors based on a common phenomenon.
This redundancy needs to be exploited by the routing protocols to improve energy and
bandwidth utilization.
Among possible classications,routing protocols can be grouped into three categories,
namely,proactive,reactive,and hybrid protocols depending on how the source nds a
route to the destination.Proactive routing protocols oer update routes to any destination
in the network at any given time by frequently updating the routing tables.The main
advantage of this approach is the minimal delay an application experiences when it wants
to send information across the network.Keeping track of all the routes through the
network,on the other hand,introduces a large protocol overhead.The routing table
updates are usually periodic but can be extended by event-driven updates to quickly
react to network changes.
Reactive routing means that routing information is not gathered in advance,but rst
when requested by an application.This saves control overhead but introduces the need
for a discovery phase each time an application needs to send data over the network and
increases the communication delay.In the route discovery phase,the network is partially
or entirely ooded with route request messages to nd the shortest path to the destination.
In reactive routing,the scalability is better because routing information is only exchanged
20 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
when needed and a node can be certain to use the most recent routing information,which
leads to a higher transmission success than with proactive routing where routes might be
outdated.Flooding an entire WNS means considerable communication eorts and the
ooding should be restricted to a particular area of interest to minimize the energy loss.
The size of a WSN,its network dynamics and energy constraints are factors in favor
of reactive routing protocols using local information.This allows the network to react
quickly to local changes or bursty trac and does not require the global transmission of
routing information.On the other hand,when information has to be transmitted quickly
from nodes to the sink (e.g.when an alarm was triggered),a predetermined path from
source to sink reduces the delay considerably.
When sensor nodes are static,it is preferable to have table driven routing protocols rather
than using reactive protocols.A signicant amount of energy is used in route discovery
and setup of reactive protocols.
Al-karaki and Kamal [Al-karaki and Kamal,2004] proposed some classications of WSN
routing protocols.Depending on the network structure,routing protocols in WSNs can be
also,divided into three kinds: at-based routing,hierarchical-based routing,and location-
based routing.In at-based routing,all nodes are typically assigned equal roles or func-
tionality.However,in hierarchical-based routing,nodes will play dierent roles in the
network.And in location-based routing,sensor positions are used to route data in the
network.In some cases,certain system parameters can be controlled in order to adapt to
the current network conditions and available energy levels.In these situations,the rout-
ing protocol is considered adaptive.Furthermore,these protocols can be classied into
multipath-based,query-based,negotiation-based,QoS-based,or coherent-based routing
techniques depending on the protocol operation.In order to streamline this survey,we
use a classication according to the network structure and protocol operation (routing
criteria).The classication is shown in Figure 2.1 redrawn from [Al-karaki and Kamal,
2004].
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 21
Figure 2.1:Routing protocols taxonomy
2.2.1 Flat routing protocols
In at-based routing protocols all sensor nodes collaborate with the same role in the rout-
ing procedures.Since the number of sensor nodes is huge therefore it is not possible to
assign a individual identier to each and every node.This leads to data-centric rout-
ing approach in which the base station sends query to a group of particular nodes in a
region and waits for response.So,in data-centric routing paradigm,data is important
than the individual nodes identities.Information is referred by using attributes of the
phenomenon.For example,the query"Give me the temperature in the region R"needs
to be disseminated to sensor nodes of a region R.At the same time,data coming fromthe
region R have to be delivered to the user that sent the query.A at network architecture
presents several advantages;it has minimal overhead to maintain the infrastructure and
the abilities for the discovery of multiple routes between communicating nodes to achieve
fault tolerance.In the following,we summarize the most known at routing protocols
and highlight their advantages and performance issues.
22 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
Figure 2.2:Flooding Implosion
Figure 2.3:Flooding overlapping
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 23
Sensor Protocols for Information via Negotiation (SPIN)
Simple protocols such as ooding and gossiping are frequently proposed to achieve in-
formation dissemination in WSNs.Flooding and Gossiping [Hedetniemi and Liestman,
1988] are among the conventional routing protcols.Flooding requires that each node
transmits a copy of its data packet to all its neighbors until the information reaches all
nodes in the network.On the other hand,gossiping uses random transmission to reduce
the number of duplicate data packets.It requires only that a node receiving a data packet
retransmits it to a randomly chosen neighbor.The simplicity of fooding and gossiping
is benec since both protocols use simple forwarding basis and do not need topology
maintenance.However,the performance of these algorithms in terms of packet delivery
delay and resource utilization,decreases quickly with the size of the network and the
trac load.The performance fall is caused by trac implosion as depicted in Figure 2.2
and geographical overlapping (Figure 2.3 redrawn from [K.Sohraby,2007]).Implosion
is due to multiple duplication of the same data being transmitted to the same sensor
node by several neighboring nodes.On the other hand,geographical overlapping makes
nodes that cover the same geographical area to spread,needlessly,identical data infor-
mation items to the network sensor nodes.Moreover,simple protocols such as fooding
and gossiping do not change their behavior to adjust communication and computation
to the current state of their energy resource.This deprivation of resource awareness and
adaptation may decrease signicantly the network lifetime,since highly active nodes may
rapidly drain their energy resources.In order to improve the classic ooding and gos-
siping,Sensor Protocols for Information via Negotiation (SPIN) is proposed.SPIN is a
data-centric negotiation-based family of information dissemination protocols for WSNs
[W.Heinzelman,1999,J.Kulik,2002].It is a family of adaptive protocols,which e-
ciently disseminate information in a WSN.The principles of this family of protocols are
the use of data negotiation and resource-adaptive algorithms.Nodes running SPIN assign
a high-level name to describe their collected data,called meta-data,and perform meta-
data negotiations before any data is transmitted between network nodes.A receiver that
indicates interest in the data content can send a request to get the advertised data.This
scheme of negotiation guarantees that data are sent only to nodes that are interested in
24 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
Figure 2.4:SPIN basic protocol operations
it,consequently,avoiding trac implosion and reducing considerably the transmission
of redundant data over the network.Besides,the use of meta data descriptors excludes
the possibility of overlap,since nodes can limit their requests to express only the data
that they are interested in getting.These negotiations assure that redundant data is not
transmitted throughout the network.SPIN uses three messages to exchange data between
nodes:The ADV message that allows a sensor to advertise a particular meta-data,the
REQ message to request the needed data and the DATA message that transfert the actual
data.Figure 2.4.,redrawn from [W.Heinzelman,1999],summarizes the function of the
SPIN protocol.In this gure,node A will transmit a data packet,it starts by advertising
its data to node B.Node B responds by sending a request to node A.After receiving the
requested data,node B then sends out advertisements to its neighbors,who in turn send
requests back to B.SPIN presents some advantages.First,any topological changes are
localized since each node needs to know only its single-hop neighbors.Second,it highly
reduces energy consumption compared to ooding.Three,it also achieves high data dis-
semination rates.However,SPIN data advertisement mechanism cannot guarantee the
delivery of data.For instance,if the nodes that are interested in the data are far away
from the source node and the nodes between source and destination are not interested in
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 25
that data,such data will not be delivered to the destination at all.Consecontly,SPIN is
not a good alternative for applications such as intrusion detection,which require reliable
periodic delivery of data packets.
Directed Diusion
Directed Diusion[C.Intanagonwiwat,2000,et al.,1999] is an important improvement in
the data-centric routing protocols for sensor networks.The goal behind Directed Diusion
is to diuse data over sensor nodes by naming it.The main reason behind using such
a scheme is to reduce unnecessary operations of network layer routing in order to save
energy.To understand the naming mecanism,in the following,we give an example of
task naming (interest):
type = wheeled vehicle//detect vehicle location
interval = 20 ms//send events every 20 ms
duration = 10 seconds//for the next 10 seconds
rect = [-100,100,200,400]//from sensors within rectangle
In this example,the sink is interested in receiving the vehicle location in the region rec
every interval for the next duration.Directed Diusion combines data coming from
dierent source and en-route them by eliminating redundancy,minimizing the number
of data transmission;thus maximizing network lifetime.The base station requests data
by broadcasting an interest message which contains a description of a sensing task.This
interest message propagates through the network hop-by-hop and each node also broadcast
interest message to its neighbor.As interest message propagates throughout the network,
gradients are setup by every node within the network.The gradient is a reply link to a
neighbor from which the interest was received.This process continues until gradients are
setup from source node to base station.Consequently,by using interest and gradients,
paths are created between sink and sources.Several paths can be established,and loops
can exist.Loops are not checked at this stage but removed at later stage.After this path of
information ow are formed and then best path are reinforced to prevent further ooding
according to a local rule.Data aggregation took place on the way of dierent paths from
dierent sources to base station or sink.To provide reliability,the base station periodically
26 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
Figure 2.5:Directed Diusion protocol operations
refresh and resend the interest message as soon as it start to receives data from sources.
Figure 2.5,redrawn from[C.Intanagonwiwat,2000],summarizes the Directed Diusion
protocol.To enhance reliability,path repairs are also possible in Directed Diusion.Thus,
when a path between a source and the sink falls,an alternative path should be selected.To
do that,Directed Diusion fundamentally regenerates reinforcement by exploring among
other paths,which are sending data in lower rates.The authors of [Ganesan et al.,2002]
proposed the use of multiple paths in advance so that in case of a failure of a path,one of
the alternative paths is chosen without any cost for searching for another one.There is
extra overhead for keeping these substitute paths alive by using low data rate,which will
clearly use extra energy but more energy can be conserved when a path fails and a new
path should be chosen.Directed Diusion and SPIN dier in terms of the on demand
data querying mechanismit has.In Directed Diusion,the base station queries the sensor
nodes if a specic data is available,by ooding some tasks.In SPIN protocols,sensors
announce the availability of data allowing interested nodes to query that data.Directed
Diusion has numerous advantages.Because of it is data centric,all transmission is
neighbor-to-neighbor with no requirement for a node addressing mechanism.Each node
can do data aggregation and caching,in addition to do environment sensing.Data caching
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 27
is an important advantage in terms of energy eciency and delay.Besides,Direct Diusion
is highly energy ecient because it is on demand and there is no need for maintaining
global network topology knowledge.On the other hand,Directed Diusion cannot be
appropriate to all sensor network applications as it is based on a query-driven data delivery
model.The applications that require continuous or periodic data delivery to the sink will
not work eciently with a query-driven on demand data model.Consequently,Directed
Diusion is not a good alternative as a routing protocol for the applications such as
environmental monitoring.Moreover,the naming schemes used in Directed Diusion are
application based and each time should be dened a priori.In addition,the matching
process for data and queries might request some extra overhead at the sensors.
Rumor routing
Rumor routing[Braginsky and Estrin,2002] is another version of Directed Diusion and
is mainly designed for applications where geographic routing is not feasible.In general,
Directed Diusion uses ooding to introduce the query to the whole network when there
is no geographic rule to diuse tasks.However,in some situations,there is only a little
amount of data that requested from the nodes and so,the use of ooding is unneeded.
An alternative method is to ood the events only if the number of events is small and the
number of queries is important.The aim is to route the queries to the nodes that have
detected a particular event rather than ooding the whole network to get information
about the occurring events.The Rumor Routing algorithm uses long-lived data packets,
called agents to ood events through the network.So,when a node observes an event,it
adds this event to its local table that called events table,and creates an agent.Agents
traverse the network in order to spread information about local events to distant nodes.
When a node creates a query for an event,the nodes knowing the route to access to that
event,may respond to the query by examining its event table.Therefore,there is no
require to ood the entire network,which diminishes the energy communication cost.On
the other hand,Rumor Routing conserves only one path between source and destination
as against to Directed Diusion where data can be transmitted through multiple paths at
lowrates.However,Rumor Routing operates well only when the number of events is small.
28 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
For a large number of events,the cost of maintaining agents and event-tables in each node
becomes impracticable if there is not enough interest in these events fromthe base station.
In addition,the overhead involved by Rumor Routing is controlled by dierent algorithm
parameters such as time-to-live (TTL) appropriate to queries and agents.Because the
nodes become aware of events through the event agents,the proceeding for selecting the
route of an event agent highly aects the performance of next hop selection in rumor
routing.
Minimum Cost Forwarding Algorithm (MCFA)
The MCFA algorithm[F.Ye,2001] exploits the idea that the direction of routing is always
known,which is towards the xed base station.Consequently,a sensor node require not
have a unique ID nor keep a routing table.As an alternative,each node maintains the
least cost estimated to transmit from itself to the base station.Then,each packet to
be forwarded by the sensor node is broadcasted to its neighbors.Receiving the packet,
the node checks if it is on the least cost path between the source sensor node and the
base station.If this node nd that is on the least cost path,it rebroadcasts the packet
to its neighbors.This method repeats until the base station is attained.In a network
using MCFA,each node must know the least cost path estimate from itself to the base
station.To do that,the base station broadcasts an advertisement message with the
cost initialized to zero while every node initially set its least cost to the base station
to innity.Upon receiving the broadcast packet created at the base station,each node
checks to see if the estimate in the packet plus the link on which it is received is less
than the current estimate.If it is the case,the current estimate and the estimate in the
advertisement message are updated.If the received broadcast message is updated,then
it is forwarded;else,it is purged and nothing further is done.As result of forwarding
advertisement message immediately after updating,some nodes will have multiple updates
and do multiple forwards as lesser cost estimates ow in.Furthermore,the nodes far
away from the base station will get more updates from those closer to the base station.
To avoid this instability during the setup phase,the MCFA was modied to run a backo
algorithm at this phase.The backo algorithm dictates that a node will not send the
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 29
updated message until A  C
node
time units have elapsed from the time at which the
message is updated,where A is a constant determined through simulation and C
node
is
the link cost from which the message was received.
Gradient-Based Routing
Schurgers et al.[C.Schurgers,2001] proposed another variant of directed diusion,called
Gradient-Based Routing (GBR).The key idea behind GBR is to keep the number of hops
when the interest is diused through the whole network.Which means that each node
can calculate a parameter called the height of the node.This parameter is the minimum
number of hops to attain the base station.The gradient on the link between a node and
its neighbor is dened as the dierence between a node height and that of its neighbor.
So that,a packet is forwarded on a link with the largest gradient.GBR protocol uses
some additional techniques such as data fusion and trac spreading in order to uniformly
balance the trac over the network.When multiple paths pass through a node,which
acts as a relay node,that relay node may aggregate data according to a certain function.
In GBR,three dierent data dissemination techniques have been proposed (1) Stochastic
Scheme,where a node chooses one gradient at random when there are two or more next
hops that have the same gradient,(2) Energy-based scheme,where a node increases its
height when its energy falls below a certain threshold,so that other sensors are prevented
from sending data to that node,and (3) Stream-based scheme,where new streams are
not relayed through nodes that are currently part of the path of other streams.The
main objective of these schemes is to achieve a balanced distribution of the trac in the
network,which allows extending the network lifetime.
Information-driven sensor querying (IDSQ) and Constrained anisotropic dif-
fusion routing (CADR)
The authors of [M.Chu,2002] proposed two routing techniques,namely,information-
driven sensor querying (IDSQ) and constrained anisotropic diusion routing (CADR).
CADR has a goal to be a general form of directed diusion.The key idea behind CADR
is to query sensors and route data in the network in order to maximize the information
30 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
gain,while minimizing latency and bandwidth.CADR diuses queries by using a set
of information criteria to choose sensors that can get the data.This is performed by
activating only the sensors that are close to a particular event and dynamically adapting
data routes.CADR takes in consideration both information gain and communication
cost,which is the main dierence from directed diusion.Thus,each node evaluates
an information/cost objective and forwards data based on the local information/cost
gradient and end-user stipulations.To evaluate information/cost,an estimation theory
was employed to model information utility measure.In IDSQ,the querying node can
determine which node can provide the most pertinent information with the additional
opportunity of distributing the energy cost.Even so,IDSQ does not precisely dene
how the query and the information are routed between sensors and the base station.
Consequently,IDSQ can be seen as a complementary optimization procedure for the
protocol.As compared to Directed Diusion,these approaches are more energy-ecient
where queries are diused in an isotropic fashion and reaching nearest neighbors rst.
COUGAR
COUGAR [Y.Yao,2002] views the network as a big distributed database system.The
principle of this protocol,is to use declarative queries to abstract query processing from
the network layer functions such as selection of appropriate sensors,etc.COUGAR uses
in-network data aggregation to save energy.The abstraction is provided through an new
query layer lying between the network and application layers.COUGAR proposes an
architecture for the sensor database system where sensor nodes choose a leader node
to achieve aggregation and transmit the data to the sink( see Figure 2.6 redrawn from
[Y.Yao,2002]).The sink takes charge of generating a query plan,which denes the
necessary information about the data ow and in-network computation for the incoming
query and transmit it to the relevant nodes.The query plan also indicates how to select
a leader for the query.The architecture provides in-network computation capacity that
can provide energy eciency in situations when the generated data is enermous.To
assure abstraction,COUGAR provided a network-layer independent methods for data
query.But,COUGAR has some drawbacks.First,the use of query layer on each sensor
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 31
Figure 2.6:the leader node gets all the readings,calculates the average and if it is greater
than a threshold sends it to the base station
node may cause an extra overhead in terms of energy consumption and memory storage.
Second,to achieve successful in-network data computation,before sending the data to the
leader node,nodes must be synchronized because not all data are received at the same
time from incoming sources.Third,to prevent nodes failure,the leader nodes should be
dynamically maintained,because leader role is high energy draining.
ACQUIRE
Sadagopan et al.[N.Sadagopan,2003],proposed a technique for gathering data from
sensor networks called ACtive QUery forwarding In sensoR nEtworks (ACQUIRE).As
COUGAR,ACQUIRE consideres the network as a distributed database.To drop with
complex queries,ACQUIRE proposes dividing each complex query into several sub-
queries.In the following,the operation of ACQUIRE is described.The BS transmits
a query to the network,This query is forwarded by each node receiving it.Each node
that receives the query,tries to respond to it partially by using its pre-stored information
and then forward it to another sensor node.If the pre-stored information is not up-to-
date,the nodes collecte information from their neighbors within a look-ahead of d hops.
32 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
Once the query is being completely resolved,it is sent back through either the reverse
or shortest-path to the sink.Consequently,ACQUIRE is suitable for complex queries by
making many nodes to send responses.Since data-centric approaches such as directed
diusion uses ooding-based query mechanism for continuous and aggregate queries so
that it may not be used for complex queries due to energy considerations.ACQUIRE can
performecient querying by adapting the value of the look-ahead parameter d.Note that
when d is equal to network diameter,ACQUIRE mechanism becomes similar to ooding.
On the other hand,if d is too small,the query has to travel more hops.Therefore,an
optimal value of d must be found,then a mathematical modeling has been proposed to
nd such value for a grid of sensors where each node has four immediate neighbors.How-
ever,there is no validation of results through simulation and the reception costs have not
taken into account during calculations.To select the next node for forwarding the query,
ACQUIRE either takes it randomly or the selection is based on maximum potential of
query fulllment.The problem of selecting the next node for forwarding the query,which
ACQUIRE addresses,is based on either information gain (CADR and IDSQ) or query
is forwarded to a node,which knows the path to the searched event (rumor routing) as
described earlier.
Energy Aware Routing
Energy-Aware Routing protocol (EAR)[R.C.Shah,2002] is a destination initiated reac-
tive protocol that the aim is to extend the network lifetime.EAR protocol is similar to
directed diusion but,it maintains a set of paths rather of maintaining or enforcing one
optimal path at higher rates.These paths are keeped and selected by means of a certain
probability.This probability depends on the manner that the low energy consumption of
each path can be achieved.Since having paths selected at dierent times,the energy of
any single path will not drain quickly,which can perform large network lifetime because
energy is consumed equitably among all nodes.Network lifetime is the main metric of
this protocol.In EAR each node is addressable through a class-based addressing which
includes the location and types of the nodes.This protocol initiates a connection through
localized ooding,which is used to nd all routes between source and destination and
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 33
their costs;thus the routing tables are built.The high-cost paths are rejected and a
forwarding table is built by selecting neighboring nodes in a fashion that is proportional
to their cost.Then,forwarding tables are used to transmit data to the destination using
paths with a probability that is inversely proportional to the node cost.To keep the paths
alive,localized ooding is performed by the destination node.As compared to directed
diusion,this protocol performs an overall improvement of 21.5% energy saving and a
44% increase in network lifetime.On the other hand,this manner requires collecting the
location information and setting up the addressing mechanism for the nodes,which make
dicult route setup compared to the directed diusion.
2.2.2 Location-based routing protocols
In this category of protocols,routing decisions are based on cached geographical infor-
mation of neighboring sensor nodes to nd a relatively optimal path without ooding
and make simplied packet forwarding decisions.This,improve network scalability by
reducing the total routing overhead.Many techniques are used to determine the node
geographical information.Global Positioning System (GPS) is the most well-known loca-
tion service in use today.Therfore,some nodes are equipped with a small low power GPS
receiver [Y.Xu,2001a] However,the GPS approach is relatively expensive for low-cost,
ad-hoc sensor networks,since GPS is based on extensive infrastructure and not available
in special environments such as indoors.Instead of the GPS use,relative coordinates of
neighboring nodes can be obtained by exchanging such information between neighbors.
The information such as connectivity,incoming signal strength,time of ight,angle of
arrival,etc.are successfully used to determine the position of sensor node with only local-
ized computations [N.Bulusu,2000,A.Savvides,2001,S.Capkun,2001,Y.Shang,2003,
Romer,2003,L.Doherty,2001].The rest of this section,reviews most of the location or
geographic based routing protocols for wireless sensor networks.
Geographic Adaptive Fidelity (GAF)
Designed primarily for mobile ad hoc networks,GAF [Y.Xu,2001b] is an energy-aware
location-based routing algorithm that may be applicable to sensor networks as well.The
34 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
Figure 2.7:Virtual grid and active nodes in GAF
whole network area is rst divided into xed zones and form a virtual grid as shown in
Figure 2.7 redrawn from [Y.Xu,2001b].Inside each grid,nodes;which are equivalent
for routing,collaborate with each other to play dierent roles.Then,nodes will elect one
sensor node to stay awake for a certain period of time and then they go to sleep.This
node is responsible for monitoring and reporting data to the Base Station on behalf of
the nodes in the zone.Hence,GAF conserves energy by turning o unnecessary nodes
in the network without aecting the level of routing delity.Each node uses its location
information to associate itself within a point in the virtual grid (for example,node uses its
GPS-indicated location or computes its location based on specic technique to determine
its location).Nodes grouped with the same point on the grid are considered equivalent in
terms of the cost of data packet routing.Such equivalence is employed to maintain some
nodes located in a particular grid area in sleeping state in order to save energy.Thus,
GAF can substantially increase the network lifetime as the number of nodes increases.To
extend the network lifetime,GAF also uses a load balancing strategy.It tries to rotate
the role of active node in each grid periodically.In high mobility cases,the leaving node
informs its neighbor in the grid about its estimated leaving time,so that neighboring
nodes can adapt their sleeping schedule to provide the changes.GAF dened three states
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 35
for nodes;discovery,for searching the neighbors in the grid,active re ecting participation
in routing and sleep when the radio is turned o.With the aimof dominating the mobility,
each node in the grid estimates its leaving time of grid and transmits this to its neighbors.
The sleeping neighbors adapt their sleeping time thus in order to maintain the routing
delity.So,sleeping nodes wake up and one of them becomes active,earlier than the
leaving time of the active node expires.Two versions of GAF were implemented both for
non-mobility (GAF-basic) and mobility (GAF-mobility adaptation) of nodes.The issue
is how to schedule roles for the nodes to act as cluster-heads.A cluster-head can query
the sensor nodes in its cluster to switch on and start gathering data if it senses an object.
Then,cluster-head is liable for receiving raw data from other nodes in its cluster and
transmit it to the BS.GAF performs at least as well as a normal ad hoc routing protocol
in terms of latency and packet loss.It also increases network lifetime proportionally to
node density.However,GAF assumes that sensor nodes know their locations using a GPS
receiver,which is impossible in the actual technology on a sensor node.
Geographic and Energy Aware Routing (GEAR)
Geographical Energy Aware Routing (GEAR) [Y.Yu,2001] is a geographic protocol
that is energy aware,to increase the lifetime of sensor networks.GEAR uses geographic
information to limit the query ooding in Direct Diusion.Rather than send interest
in entire network as done in Directed diusion,in GEAR,the interest is only sent to
a certain region.To select neighbors,GEAR uses energy aware metrics in such a way
that each node attempts to balance the energy consumption among its neighbors.An
estimation cost;which is a combination of residual energy and distance to the destination
through its neighbors,is maintained in each node.The node also maintains a learned
cost,which is a renement of the estimated cost that accounts for routing around holes in
the network.In case when no other neighbors are closer to the destination than itself,a
node confronts a routing hole.Consequently,the learned cost is larger than the estimated
cost.The estimated cost is adapted when the learned cost is sent one hop back every time
a packet reaches the destination.There are two phases in GEAR algorithm:First,when
transmitting packets towards the target region,a node examines its neighbors to nd any
36 CHAPTER 2.WIRELESS SENSOR NETWORKS ROUTING PROTOCOLS
one,which is closer to the target region than itself.In the case when there is more than
one,the nearest neighbor to the target region is chosen for the next hop.There is a
hole if they are all further than the node itself.In this situation,one of the neighbors is
taken to forward the packet based on the learning cost function.This choice can then be
updated in relation to the convergence of the learned cost during the delivery of packets.
Second,if the packet has attained the target,it can be diused in that region by either
recursive geographic forwarding or restricted ooding.Restricted ooding can be used to
forward the data to the destination in case where the target region node density is low.
However,when the target region node density is high,recursive geographic ooding can
be used to save energy by dividing the target region into sub regions and sending one copy
to each sub-region.This process continues until the destination is reached.Compared
to a similar non-energy-aware routing protocols,GEAR reduces energy consumption and
performs better in terms of packet delivery.However,GEAR is sensitive to location errors
and it increases the average path length for a packet.Even if GEAR decreases the number
of states a node should maintain,it requires a location service to map locations and node
identiers,which is considered too exorbitant for wireless sensor networks.
2.2.3 Hierarchial routing protocols
In a densely deployed sensor network,the physical environment would generate very sim-
ilar data in close-by sensor nodes and transmitting such data is more or less redundant.
An event (an intruder,a change in temperature,a chemical substance) is often detected
by more than one sensor and duplicated data is generated.From a QoS point of view,
this redundancy can be important since it will be used to compensate the lost packets
or failed links.However,this redundancy is often eliminated,which can not only di-
minish the global data to be transmitted and localized most trac to within individual
groups,but reduces the trac and consequently,contention in a wireless sensor network.
A way to reduce energy consumption is data aggregation,which consists of suppressing
redundancy in dierent data messages.This data aggregation is the key idea for the
most Hierarchical routing protocols.In addition,scalability is one of the major design
attributes of sensor networks.A single-tier network can lead the gateway to overload
2.2.ROUTING PROTOCOLS FOR WIRELESS SENSOR NETWORKS 37
with the increase in sensors density.Such overload might provoke latency in communi-
cation and unacceptable tracking of events.Also,the single-gateway architecture is not
scalable for a larger set of sensors covering a large area of interest because the sensors are
typically not capable of long distance communication.To allow the system to deal with
additional load and to be able to cover a large area of interest without deteriorating the
service,networking clustering has been addressed in some routing approaches.The main
target of hierarchical routing is to eciently maintain the energy consumption of sensor