"EDDEEC", "BEENISH" and "i-BEENISH" Energy Efficient Routing Protocols for Heterogeneous WSNs

Abstract

In past years there has been increasing interest in field of Wireless Sensor Networks (WSNs). WSNs consist of large number of randomly distributed energy constrained sensor nodes. Sensor nodes have ability to sense and send sensed data to Base Station (BS). Sensing as well as transmitting data towards BS require high energy. Saving energy and extending network lifetime are great challenges. One of the major issue in WSN is development of energy efficient routing protocols. Clustering is an effective way to increase energy efficiency. Mostly heterogenous WSN protocols consider two or three energy level of nodes. In this thesis, I propose multiple novel clustering based routing techniques: Enhanced Developed Distributed Energy Efficient Clustering (EDDEEC), BEENISH (Balanced Energy Efficient Network Integrated Super Heterogenous) and Improved BEENISH (i-BEENISH) for heterogeneous WSNs. EDDEEC scheme is based on dynamically changing Cluster Head (CH) selection chances with more efficiency in three level heterogeneous WSN. BEENISH and i-BEENISH consider four types of nodes; normal, advanced, super and ultra-super nodes based on their initial energy in WSN. In BEENISH, CHs selection is made on residual energy level basis of nodes. Where as, i-BEENISH adaptively and more efficiently changes the CH election probability of high energy nodes when their energy get lower. Finally, the simulation results show that EDDEEC, BEENISH and i-BEENISH perform better than current important clustering protocols in heterogeneous WSNs containing varying heterogeneity level. The proposed protocols achieve longer stability, lifetime and more effective messages than Distributed Energy Efficient Clustering (DEEC), Developed DEEC (DDEEC) and Enhanced DEEC (EDEEC).

Full text

i
“EDDEEC”, “BEENISH” and “i-BEENISH”
Energy Efficient Routing Protocols for
Heterogeneous WSNs
By
Mr. Muhammad Talha Naeem Qureshi
Registration Number: CIIT/FA11-REE-001/ISB
MS Thesis
In
Electrical Engineering
COMSATS Institute of Information Technology
Islamabad – Pakistan
FALL, 2012

ii
“EDDEEC”, “BEENISH” and “i-BEENISH” Energy
Efficient Routing Protocols for Heterogeneous
WSNs
A Thesis presented to
COMSATS Institute of Information Technology
In partial fulfillment
of the requirement for the degree of
MS (Electrical Engineering)
By
Mr. Muhammad Talha Naeem Qureshi
CIIT/FA11-REE-001/ISB
Fall, 2012

iii
“EDDEEC”, “BEENISH” and “i-BEENISH” Energy
Efficient Routing Protocols for Heterogeneous
WSNs
A Graduate Thesis submitted to Department of Electrical Engineering as
partial fulfillment of the requirement for the award of Degree of M. S.
(Electrical Engineering).
Name
Registration Number
Mr. Muhammad Talha Naeem Qureshi
CIIT/FA11-REE-001/ISB
Supervisor:
Dr. Nadeem Javaid,
Assistant Professor,
Center for Advanced Studies in Telecommunications (CAST),
COMSATS Institute of Information Technology (CIIT),
Islamabad Campus,
December, 2012

iv
Final Approval
This thesis titled
“EDDEEC”, “BEENISH” and “i-BEENISH” Energy
Efficient Routing Protocols for Heterogeneous
WSNs
By
Mr. Muhammad Talha Naeem Qureshi
CIIT/FA11-REE-001/ISB
has been approved
for the COMSATS Institute of Information Technology, Islamabad
External Examiner: __________________________________
(To be decided)
Supervisor: ________________________
Dr. Nadeem Javaid /Assistant professor,
Center for Advanced Studies in Telecommunications (CAST),
CIIT, Islamabad.
Head of Department:________________________
Dr. Raja Ali Riaz / Associate professor,
Department of Electrical Engineering,
CIIT, Islamabad.

v
Declaration
I Mr. Muhammad Talha Naeem Qureshi, CIIT/FA11-REE-001/ISB here
byxdeclare that I have produced the work presented inxthis thesis,
duringxthe scheduledxperiod of study. I also declare that I havexnot taken
anyxmaterial from anyxsource exceptxreferred toxwherever due that
amountxof plagiarism isxwithin acceptablexrange. If a violationxof HEC
rulesxon research hasxoccurred in thisxthesis, I shall be liablexto
punishablexaction under the plagiarismxrules of the HEC.
Date: ________________
____________________________
Muhammad Talha Naeem Qureshi
CIIT/FA11-REE-001/ISB

vi
Certificate
It is certified that Mr. Muhammad Talha Naeem Qureshi, CIIT/FA11-REE-
001/ISB has carried out all the work related to this thesis under my
supervision at the Department of Electrical Engineering, COMSATS
Institute of Information Technology, Islamabad and the work fulfills the
requirements for the award of MS degree.
Date: _________________
Supervisor:____________________
Dr. Nadeem Javaid /Assistant professor,
Center for Advanced Studies in Telecommunications (CAST),
CIIT, Islamabad.
____________________________
Head of Department:
Dr. Raja Ali Riaz/Associate Professor,
Department of Electrical Engineering,
CIIT, Islamabad.

vii
DEDICATION
Dedicated to my family.

viii
ACKNOWLEDGMENT
I am heartily grateful to my supervisor, Dr. Nadeem Javaid whose patient
encouragement, guidance and insightful criticism from the beginning to the
final level enabled me have a deep understanding of the thesis.
Lastly, I offer my profound regard and blessing to everyone who supported
me in any respect during the completion of my thesis especially my friends
in every way offered much assistance before, during and at completion
stage of this thesis work.
Mr. Muhammad Talha Naeem Qureshi
CIIT/FA11-REE-001/ISB

ix
List of Abbreviations
WSNs
Wireless Sensor Networks
LEACH
Low Energy Adaptive Clustering Hierarchy
MEMS
Micro Electro Mechanical Sensor
EDDEEC
Enhanced Developed Distributed Energy Efficient Clustering
SEP
Stable Election Probability
DEEC
Distributed Energy Efficient Clustering
BEENISH
Balanced Energy Efficient Network Integrated Super Heterogeneous
i-BEENISH
Improved Balanced Energy Efficient Network Integrated Super
Heterogeneous
CH
Cluster Head
BS
Base Station
DDEEC
Developed Distributed Energy Efficient Clustering
EDEEC
Enhanced Distributed Energy Efficient Clustering
PEGASIS
Power Efficient Gathering in Sensor Information Systems
HEED
Hybrid Energy Efficient Distributed Clustering
DC
Direct Communication
HEED
Hybrid Energy Efficient Distributed Clustering

x
List of Publications
[1] Qureshi. T. N, Javaid. N, Malik. M, Qasim. U, Khan. Z. A, On Performance Evaluation of
Variants of DEEC in WSNs, published in 7th International Conference on Broadband and
Wireless Computing, Communication and Applications (BWCCA-2012), Victoria, Canada,
2012.
[2] Tauseef Shah, Nadeem Javaid, Talha Naeem Qureshi, “Energy Efficient Sleep Awake Aware
(EESAA) Intelligent Sensor Network Routing Protocol”, published in 15th IEEE International
Multi Topic Conference (INMIC’12), 2012, Pakistan.
[3] T. N. Qureshi, N. Javaid, Z. A. Khan, " Enhanced Developed Distributed Energy-Efficient
Clustering (EDDEEC) for Heterogeneous Wireless Sensor Networks ", submitted in, 10th IEEE
International Conference on Wireless On-demand Network Systems and Services (WONS'13),
March 18-20, 2013, Banff, Canada.
[4] T. N. Qureshi, N. Javaid, Z. A. Khan, "BEENISH: Balanced Energy Efficient Network
Integrated Super Heterogenous Protocol for Wireless Sensor Networks", submitted in, 10th IEEE
International Conference on Wireless On-demand Network Systems and Services (WONS'13),
March 18-20, 2013, Banff, Canada.
[5] T. N. Qureshi, N. Javaid, Z. A. Khan, " ABEENISH: Adaptive Balanced Energy Efficient
Network Integrated Super Heterogenous Protocol for Wireless Sensor Networks ", submitted in
4th IEEE International Conference on Ambient Systems, Networks and Technologies (ANT-13),
June 25-28, 2013, Halifax, Nova Scotia, Canada.

Table of Contents
1 Abstract 1
2 Introduction 2
2.1 Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . . . . 2
2.2 RoutinginWSN ............................ 2
2.3 Centralized and Distributed Algorithms . . . . . . . . . . . . . . . . 3
2.4 ClusteringinWSN ........................... 3
2.5 Homogenous and Heterogenous WSN . . . . . . . . . . . . . . . . . 3
3 Background and Motivation for Thesis 6
3.1 Background ............................... 6
3.2 Motivation................................ 7
4 Heterogeneous WSN Models 9
4.0.1 Two Level Heterogeneous WSN Model . . . . . . . . . . . . 9
4.0.2 Three Level Heterogeneous WSN Model . . . . . . . . . . . 9
4.0.3 Four Level Heterogeneous WSN Model . . . . . . . . . . . . 10
4.0.4 Multi-level Heterogeneous WSN Model . . . . . . . . . . . . 11
4.1 Radio Dissipation Model . . . . . . . . . . . . . . . . . . . . . . . . 11
5 EDDEEC Protocol 13
6 BEENISH and i-BEENISH Protocols 17
6.1 BEENISHProtocol........................... 17
6.2 i-BEENISHProtocol .......................... 19
6.3 Performance Criteria Used . . . . . . . . . . . . . . . . . . . . . . . 22
7 Simulation Results 24
7.1 Simulation Results of EDDEEC . . . . . . . . . . . . . . . . . . . . 24
7.1.1 Case 1: m= 0.5, mo= 0.4, a = 1.5and b =3 ........ 24
7.1.2 Case 2: m= 0.8, mo= 0.6, a = 2.0and b = 3.5 ....... 25
7.1.3 Case 3: m= 0.3, mo= 0.2, a = 1.2and b = 2.5 ....... 27
xi

7.1.4 Case 4: Multi-level Heterogenity . . . . . . . . . . . . . . . . 28
7.2 Simulation Results of BEENISH and i-BEENISH . . . . . . . . . . 31
7.2.1 Case 5: m= 0.5, mo= 0.3, m1= 0.2, a = 1.5, b =
2.0and u = 2.5 ......................... 31
7.2.2 Case 6: m= 0.6, mo= 0.5, m1= 0.3, a = 2.0, b =
2.5and u = 3.0 ......................... 31
7.2.3 Case 7: Multi-level Heterogeneity . . . . . . . . . . . . . . . 33
7.2.4 Case 8: m= 0.5, mo= 0.3, m1= 0.2, a = 1.5, b =
2.0and u = 2.5 ......................... 35
7.2.5 Case 9: m= 0.6, mo= 0.5, m1= 0.3, a = 2.0, b =
2.5and u = 3.0 ......................... 37
7.2.6 Case 10: Multi-level Heterogeneity . . . . . . . . . . . . . . 39
7.2.7 Case 11: m= 0.5, mo= 0.3, m1= 0.2, a = 1.5, b =
2.0and u = 2.5 ......................... 39
7.2.8 Case 12: m= 0.6, mo= 0.5, m1= 0.3, a = 2.0, b =
2.5and u = 3.0 ......................... 41
7.2.9 Case 13: Multi-level Heterogeneity . . . . . . . . . . . . . . 43
8 Conclusion 47
xii

List of Figures
3.1 Flowchart of DEEC, DDEEC, EDEEC, EDDEEC and BEENISH . 7
4.1 Radio Energy Dissipation Model . . . . . . . . . . . . . . . . . . . . 11
5.1 Round First Node dies when c is varying . . . . . . . . . . . . . . . 15
5.2 Network Topology of EDDEEC . . . . . . . . . . . . . . . . . . . . 16
6.1 Round First Node dies when z is varying . . . . . . . . . . . . . . . 21
6.2 Round First Node dies when c is varying . . . . . . . . . . . . . . . 22
6.3 Network Topology of BEENISH and i-BEENISH . . . . . . . . . . . 22
7.1 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 25
7.2 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 25
7.3 Packets Sent to Base Station . . . . . . . . . . . . . . . . . . . . . . 26
7.4 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 26
7.5 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 27
7.6 Packets Sent to Base Station . . . . . . . . . . . . . . . . . . . . . . 27
7.7 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 28
7.8 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 28
7.9 Packets Sent to Base Station . . . . . . . . . . . . . . . . . . . . . . 29
7.10 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 29
7.11 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 30
7.12 Packets Sent to Base Station . . . . . . . . . . . . . . . . . . . . . . 30
7.13 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 32
7.14 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 32
7.15PacketsSenttoBS ........................... 33
7.16 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 33
7.17 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 34
7.18PacketsSenttoBS ........................... 34
7.19 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 35
7.20 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 35
7.21PacketsSenttoBS ........................... 36
xiii

7.22 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 36
7.23 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 37
7.24PacketsSenttoBS ........................... 37
7.25 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 38
7.26 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 38
7.27PacketsSenttoBS ........................... 39
7.28 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 40
7.29 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 40
7.30PacketsSenttoBS ........................... 41
7.31 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 41
7.32 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 42
7.33PacketsSenttoBS ........................... 42
7.34 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 43
7.35 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 43
7.36PacketsSenttoBS ........................... 44
7.37 Nodes Dead During Network Lifetime . . . . . . . . . . . . . . . . . 44
7.38 Nodes Alive During Network Lifetime . . . . . . . . . . . . . . . . . 45
7.39PacketsSenttoBS ........................... 45
xiv

List of Tables
6.1 Simulation Parameters . . . . . . . . . . . . . . . . . . . . . . . . . 21
7.1 Comparing Performance of DEEC, DDEEC, EDEEC and EDDEEC
for m= 0.5, mo= 0.4, a = 1.5and b =3............... 44
7.2 Comparing Performance of DEEC, DDEEC, EDEEC and EDDEEC
for m= 0.8, mo= 0.6, a = 2.0and b = 3.5.............. 45
7.3 Comparing Performance of DEEC, DDEEC, EDEEC and EDDEEC
for m= 0.3, mo= 0.2, a = 1.2and b = 2.5.............. 46
7.4 Comparing Performance of DEEC, DDEEC, EDEEC and EDDEEC
for Multi-level Heterogenity . . . . . . . . . . . . . . . . . . . . . . 46
7.5 Comparing Performance of DEEC, DDEEC, EDEEC BEENISH
and i-BEENISH for m= 0.5, mo= 0.3, m1= 0.2, a = 1.5, b =
2.0and u = 2.5 ............................. 46
7.6 Comparing Performance of DEEC, DDEEC, EDEEC BEENISH
and i-BEENISH for m= 0.6, mo= 0.5, m1= 0.3, a = 2.0, b =
2.5and u = 3.0 ............................. 46
7.7 Comparing Performance of DEEC, DDEEC, EDEEC BEENISH
and i-BEENISH for Multi-level Heterogenity . . . . . . . . . . . . . 46
xv

Chapter 1
Abstract
In past years there has been increasing interest in field of Wireless Sensor Net-
works (WSNs). WSNs consist of large number of randomly distributed energy
constrained sensor nodes. Sensor nodes have ability to sense and send sensed data
to Base Station (BS). Sensing as well as transmitting data towards BS require high
energy. Saving energy and extending network lifetime are great challenges. One of
the major issue in WSN is development of energy efficient routing protocols. Clus-
tering is an effective way to increase energy efficiency. Mostly heterogenous WSN
protocols consider two or three energy level of nodes. In this thesis, I propose mul-
tiple novel clustering based routing techniques: Enhanced Developed Distributed
Energy Efficient Clustering (EDDEEC), BEENISH (Balanced Energy Efficient
Network Integrated Super Heterogenous) and Improved BEENISH (i-BEENISH)
for heterogeneous WSNs. EDDEEC scheme is based on dynamically changing
Cluster Head (CH) selection chances with more efficiency in three level hetero-
geneous WSN. BEENISH and i-BEENISH consider four types of nodes; normal,
advanced, super and ultra-super nodes based on their initial energy in WSN. In
BEENISH, CHs selection is made on residual energy level basis of nodes. Where
as, i-BEENISH adaptively and more efficiently changes the CH election proba-
bility of high energy nodes when their energy get lower. Finally, the simulation
results show that EDDEEC, BEENISH and i-BEENISH perform better than cur-
rent important clustering protocols in heterogeneous WSNs containing varying
heterogeneity level. The proposed protocols achieve longer stability, lifetime and
more effective messages than Distributed Energy Efficient Clustering (DEEC),
Developed DEEC (DDEEC) and Enhanced DEEC (EDEEC).
keywords
Cluster, head, residual, energy, heterogeneity, efficient, wireless, sensor, network.
1

Chapter 2
Introduction
2.1 Wireless Sensor Networks
It has become possible due to technological advancement in field of Micro Electri-
cal Mechanical Sensors (MEMS) that I have got low cost sensors that are small
in size and consume less energy. These sensors have limited processing, wireless
communication and power resource capabilities [1, 2, 3]. During past few years
Wireless Sensor Networks (WSNs) have become famous in many applications in-
cluding military surveillance, environmental, transportation traffic, temperature,
pressure and vibration monitoring. To achieve fault tolerance, WSNs contain
hundreds or even thousands of sensors randomly distributed with in the region
[4, 5, 6]. All nodes report sensed data to Base Station (BS) often termed as sink.
Nodes in WSNs are power constrained usually due to limited battery resource.
Recharging or battery replacement of already deployed sensor nodes is also not
possible because sensor nodes might be located where they can not be reached.
Sensor nodes are mostly deployed harsh working locations.
2.2 Routing in WSN
Routing protocols play important role in achieving energy efficiency in WSNs.
Usually nodes are be placed far away from BS. Therefore, frequent and long dis-
tance transmissions should be avoided to prolong network lifetime [1]. Sensor
nodes take self decisions to accomplish sensing tasks, constructing network topol-
ogy and routing policy. Therefore, it become important to design energy efficient
algorithm for enhancing robustness against node failures and extending lifetime
of WSNs.
2

2.3 Centralized and Distributed Algorithms
WSNs are usually exposed to inconsistent and harsh inconsistent environments.
Connectivity loss between nodes are probable. Centralized algorithms operate
with the complete knowledge of whole network. Transmission error or critical node
failure will possibly result in serious protocol issue [7]. Where as, distributed algo-
rithms are only executed locally without the global knowledge of whole topology.
So, distributed algorithms are robust against transmission errors, node failures
and more scalable as compare to centralized algorithms [8].
2.4 Clustering in WSN
The performance of protocols highly depends on its internal architecture. Where
protocols based on clustering architecture has good efficiency as compared to pro-
tocols based upon geographical architecture. Grouping sensor nodes in form of
clusters is beneficial in minimizing the energy utilization. Numerous energy effi-
cient protocols are made based on clustering structure [1, 9, 10]. In clustering,
nodes assemble themselves in form of clusters, one node act as the Cluster Head
(CH). All cluster member nodes transmit sensed data to their CHs, while the CH
aggregate data received and forward it to the remote BS [11, 12]. Distributed
algorithms works efficiently in clusters and do not need to wait for control mes-
sages broadcasting through the whole network. So, distributed algorithms results
in better scalability in congested networks then centralized algorithms.
2.5 Homogenous and Heterogenous WSN
Clustering can be formed in two kind of networks i.e., homogenous and heteroge-
neous. WSNs having nodes of same energy level are called homogenous WSNs.
Low Energy Adaptive Clustering Hierarchy (LEACH) [13], Power Efficient Gath-
ering in Sensor Information Systems (PEGASIS) [14] and Hybrid Energy-Efficient
Distributed Clustering (HEED) [15] are some cluster based protocols which are
designed for homogenous WSNs. These algorithms poorly perform in heteroge-
neous regions. Nodes have less energy will expire faster than high energy nodes
because these homogenous clustering based algorithms are incapable to treat ev-
ery node with respect to energy. In heterogeneous WSNs, nodes are deployed
with different initial energy levels. Heterogeneity in WSN may be the result of re
energizing of WSN in way to increase the network life span [16, 17, 18]. Stable
3

Election Protocol (SEP) [16] considers two kinds of nodes with respect to their
initial energy. The advanced nodes have more energy than normal nodes at the
start of the WSN. SEP increases stability period of the WSN, till the first node
dead. Distributed Energy Efficient Clustering (DEEC) [19] and Developed DEEC
(DDEEC) [20] both are designed for WSNs containing two kinds of nodes with
respect to their initial energy and both prolongs the stability period. Enhanced
DEEC (EDEEC) [21] are designed for three level heterogeneous WSNs, which is
comprised of three kinds of nodes with respect to initial energy. The super nodes
are provided with extra energy than advanced and normal nodes at initial stage.
In this thesis, I purpose and evaluate Enhanced Developed Distributed Energy
Efficient Clustering scheme (EDDEEC), BEENISH (Balanced Energy Efficient
Network Integrated Super Heterogenous) and Improved BEENISH (i-BEENISH)
protocols for heterogeneous WSNs. EDDEEC protocol uses the concept of three
kinds of nodes: normal, advanced and super nodes. The EDDEEC, permits to
balance the CH selection in overall nodes according to remaining energy. Super
and advanced nodes have more energy than normal nodes. So, the super and
advanced nodes are mostly selected as CH for the initial rounds and when their
most of energy exhausted to large extent and they have same energy level as of
normal nodes, these nodes will have similar CH selection criteria as normal nodes.
Therefore, energy is efficiently distributed over the network. EDDEEC extend
the network lifetime importantly the stability period by heterogeneous sensitive
clustering algorithm.
In reality, WSNs have large number of energy levels of nodes than two or three
types. Due to random selection of CH, large range of energy levels are created in
WSNs. So, as much more energy levels I can quantize will lead to as much better
results. BEENISH introduced the concept of four types of nodes; normal, ad-
vanced, super and ultra-super nodes based on their initial energy in WSN. Where
normal nodes have least initial energy level and ultra-super nodes have highest
initial energy level. BEENISH Follows the thoughts of LEACH and DEEC by
changing CH role between all nodes. In BEENISH, CH are chosen by a probabil-
ity depend on the fraction of residual energy of node and average energy of whole
network. BEENISH choose different epoch for different nodes based on their re-
maining energy. Higher energy nodes are more often elected as CH as compare to
low energy nodes. In some cases ultra-super, super and advanced nodes are extra
punished than normal nodes in BEENISH, so, i-BEENISH dynamically changes
the CH election probabilities high energy nodes when their energy get lowered due
to CH selection. Simulations show that EDDEEC, BEENISH and i-BEENISH
achieves larger stability period, network lifetime by heterogeneous aware cluster-
4

ing algorithm and more effective messages sent to BS than DEEC, DDEEC and
EDEEC in heterogeneous environments.
5

Chapter 3
Background and Motivation for
Thesis
3.1 Background
As mentioned before, there are two categories of clustering mechanisms referred
to as homogeneous and heterogeneous clustering schemes.
A clustering algorithm is introduced by Heinzeman, et. al. [13] for homogeneous
WSNs called as LEACH. In LEACH, CH role is distributed among all nodes with
in a cluster.
In PEGASIS [14], nodes are organized in form of chain. Complete topology infor-
mation is required which make this algorithm hard to apply.
HEED [15] is a distributed clustering algorithm, stochastically chooses CH. Elec-
tion criteria of every node is correlated to its remaining energy. Low energy nodes
have more election possibility than high energy nodes in HEED.
G. Smaragdakis, et. al. proposed a protocol called as SEP [16] which considers
two energy levels of nodes in WSN. CH is chosen based on residual energy of
nodes.
L .Qing, Q. Zhu, M. Wang proposed DEEC [19], in DEEC ratio of remaining
energy of node and average energy of the network is used for CH selection.
DDEEC [20] uses residual energy for CH selection. More efficient criteria for CH
selection. Advanced nodes are selected as CH for the starting rounds and when
their energy decreases, these nodes will have the same CH selection criteria like
the normal nodes.
6

Parul Saini, Ajay K. Sharma proposed EDEEC [21], extends heterogeneity to three
levels by adding super nodes.
Start Calculate
alive nodes
Calculate Eiand
Er
of every alive
node
Calculate
average energy
of the network
at present round
Calculate the
probability of
each node to
become CH
based on
residual energy
of node and
average energy
of network
Node has not been a
cluster head in previous
rounds
Yes
Random number
chosen is less than
threshold fraction
Yes
End
Node is cluster
member and
send data to
their
appropriate
cluster head
Node belongs to set G, where G
is set of nodes eligible to become
a CH and node chose a random
number between 0 and 1
Node is CH for the current round
No
No
Receving energy of
node >0.7 * initial energy
of normal node
Calculate
probability to
become CH
Modify
probabilit
y of node
to become
CH
No
Yes
Calculate probability
of each node to
become CH for
advance, normal and
super nodes, to which
ever type it belongs
and then calculate its
probability using
specific probability
function according to
its type
Assign
probabilities
depending on
its three type
DEEC
DDEEC
EDDEEC EDEEC
Receving energy of
node >0.7 * initial energy
of normal node
Yes
Modify
the
probability
of node
based on
Tabsolute
No
Calculate
CH
percentage
BEENISH
Calculate
probability of
each node to
become CH for
advance, normal
and super nodes,
to which ever
type it belongs
and then
calculate its
probability using
specific
probability
function
according to its
type
Figure 3.1: Flowchart of DEEC, DDEEC, EDEEC, EDDEEC and BEENISH
3.2 Motivation
SEP, DEEC and DDEEC consider two energy levels of nodes. EDEEC consider
three energy levels in WSN. However, CH election criteria for EDEEC is not
7

efficient. EDEEC always punishes the super and advanced nodes largely than
normal nodes, especially when their remaining energy depletes and become same
as of normal nodes. In this situation, super and advanced nodes die earlier than
the others. EDDEEC, permits to balance the CH selection overall network nodes
following their remaining energy. In reality, WSNs have large number of energy
levels of nodes than two or three types. Due to random selection of CH, large range
of energy levels are created in WSNs. So, as much more energy levels I can quantize
will lead to as much better results. BEENISH and i-BEENISH introduces concept
of four energy levels according to their initial energy. i-BEENISH adaptively
changes the CH election probability of high energy nodes when their energy get
lowered in four level heterogenous WSN.
8

Chapter 4
Heterogeneous WSN Models
WSNs contain same or different energy levels of nodes and I term it as homoge-
neous and heterogeneous WSNs, respectively. Heterogeneous WSNs may contain
two, three or multi energy levels of nodes are called as two, three and multi-level
heterogeneous WSNs, respectively.
4.0.1 Two Level Heterogeneous WSN Model
Two kinds of nodes are present. Normal nodes having energy Eoand Eo(1 + a) is
energy level of advanced nodes containing atimes increase in energy than normal
nodes. Nis the total nodes and N m is the quantity of advanced nodes. mis the
fraction of advanced nodes. The total initial energy is given by Eq. 4.1 as in [16].
Etotal =N(1 −m)Eo+Nm(1 + a)Eo
=NEo(1 −m+m+am)
=NEo(1 + am)
(4.1)
Two level heterogeneous WSNs contains am times more energy as compared to
homogeneous WSNs.
4.0.2 Three Level Heterogeneous WSN Model
Three different energy levels of nodes; normal, advanced and super nodes in net-
work. mois the fraction of super nodes having btimes increase in energy than
normal nodes. Total starting energy of three level heterogeneous WSN is given by
Eq. 4.2 and 4.3 as supposed in [21].
9

Etotal =N(1 −m)Eo+Nm(1 −mo)(1 + a)Eo+N mmoEo(1 + b) (4.2)
Etotal =NEo(1 + m(a+mob)) (4.3)
Three level heterogeneous WSNs contains m(a+mob) times more energy as com-
pared to homogeneous WSNs.
4.0.3 Four Level Heterogeneous WSN Model
I am proposing four level heterogeneous WSNs containing four different energy
levels of nodes; normal, advanced, super and ultra-super-super nodes. Hetero-
geneity may be result of re-energizing the WSN by adding some new sensor nodes
to increase lifetime of already deployed WSN [16]. Normal nodes contain energy
of Eo. Energy of advanced nodes of mfraction having atimes extra energy than
normal nodes is equal to Eo(1 + a). Whereas, energy of super nodes of fraction
mohaving btimes extra energy than normal nodes is equal to Eo(1 + b) and en-
ergy of ultra-super-super nodes of fraction m1having utimes more energy than
normal nodes is equal to Eo(1 + u). As N is total nodes in the network, then
Nmmom1is the total number of ultra-super-super nodes, Nmmo(1 −m1) is the
total number of super nodes, N m(1 −mo) is the total number of advanced nodes
and N(1 −m) is the total number of normal nodes. Total starting energy of three
level heterogeneous WSN is therefore given by Eq. 4.4 and Eq. 4.5.
Etotal =Nmmom1Eo(1 + u) + N mmo(1 −m1)Eo(1 + b)
+Nm(1 −mo)Eo(1 + a) + N(1 −m)Eo
(4.4)
Etotal =NEo(1 + m(a+mo(−a+b+m1(−b+u)))) (4.5)
The four level heterogeneous WSNs contains m(a+mo(−a+b+m1(−b+u)))
times more energy as compared to homogeneous WSNs.
10

4.0.4 Multi-level Heterogeneous WSN Model
Network containing nodes of different energy levels. Initial energy of nodes is
distributed over the close set [Eo, Eo(1 + amax)]. Total starting energy of multi-
level heterogeneous WSN is given by Eq. 4.6:
Etotal =
N
∑
i=1
Eo(1 + ai) = Eo(N+
N
∑
i=1
ai) (4.6)
A homogeneous WSNs also becomes heterogeneous after some rounds due to ran-
dom CH selection of nodes. CH nodes uses more energy, then cluster members.
After few rounds, all nodes have different energy level as compared to each other.
Therefore, a protocol which handles heterogeneity is more important than ho-
mogenous protocol.
4.1 Radio Dissipation Model
The radio energy model describes that l-bit message is transmitted over a distance
das in [9, 10] as show in Fig. 4.1. The energy expended is given in Eq. 4.7.
Figure 4.1: Radio Energy Dissipation Model
ET x(l, d) = 


lEelec +lεf s d2, d < do
lEelec +lεmp d4, d ≥do
(4.7)
Eelec is energy dissipated per bit, distance between sender and receiver is d. Free
space (fs) model is used if distance is in less than threshold otherwise multi path
(mp) model. Total dissipated energy in a round as supposed in [9, 10] is given in
Eq. 4.8 below:
Eround =L(2N Eelec +N ED A +kεmpd4
toBS +N εf s d2
toCH ) (4.8)
11

Here, K= Number of clusters,
EDA= Data aggregation energy expended in CH
dtoBS = Distance between CH and BS
dtoCH = Distance between cluster members and CH In [9, 10] it is suppoesed that
all nodes are equally distributed over network therefore, dtoBS and dtoC H can be
calculated by Eq. 4.9.
dtoCH =M
√2πk , dtoBS = 0.765M
2(4.9)
Through finding the derivative of ERound with respect to kto zero, I get the kopt
optimal number clusters as in Eq. 4.10.
kopt =√N
√2π√εfs
εmp
M
d2
toBS
(4.10)
12

Chapter 5
EDDEEC Protocol
EDDEEC protocol implements the same idea of probabilities for CH selection
based on initial, residual energy level of the nodes and average energy of network
as supposed in DEEC. However, in DEEC, probability for CH selection is based
on two kinds of nodes; normal and advanced nodes as given in Eq. 5.1.
pi=


poptEi(r)
(1+am)¯
E(r)if siis the normal node
popt(1+a)Ei(r)
(1+am)¯
E(r)if siis the advanced node
(5.1)
The average energy of rth round from [19] is given as:
¯
E(r) = 1
NEtotal(1 −r
R) (5.2)
Rrepresents total rounds during network life span and can be estimated from [19]
as:
R=Etotal
Eround
(5.3)
At start of every round, node decides whether to become a CH or not on basis of
threshold calculated by Eq. 5.4 and as supposed in [13, 19].
T(si) = 


pi
1−pi(rmod 1
Pi)if siϵ G
0otherwise
(5.4)
where, Gis the set of nodes eligible to become CH for round rand pis the
desired percentage of CH. In real scenarios, WSNs have more than two types of
heterogeneity. Therefore, in EDDEEC, I use concept of three levels heterogeneity
and characterized the nodes as normal, advanced and super nodes as supposed in
13

[21]. The probability for three types of nodes is given below:
pi=








poptEi(r)
(1+m(a+mob)) ¯
E(r)if siis the normal node
popt(1+a)Ei(r)
(1+m(a+mob)) ¯
E(r)if siis the advanced node
popt(1+b)Ei(r)
(1+m(a+mob)) ¯
E(r)if siis the super node
(5.5)
The difference between DEEC, DDEEC, EDEEC and EDDEEC is generalized in
Eq. 5.5 which defines probabilities to become CH for current round. Objective
of this expression is to distribute energy consumption over network efficiently,
increase stability period and lifetime of network. However, after some rounds, some
super and advanced nodes have same residual energy level as normal nodes due to
repeatedly CH selection. Although DEEC continues to punish just advanced nodes
whereas, EDEEC continues to punish advanced and super nodes and DDEEC is
only effective for two level heterogenous network as mentioned previously in related
work. To avoid this unbalanced case in three level heterogenous network, I propose
changes in function which defines probabilities for normal, advanced and super
nodes. These changes are based on absolute residual energy level Tabsolute, which
is the value in which advanced and super nodes have same energy level as that
of normal nodes. The idea specifies that under Tabsolute all normal, advanced and
super nodes have same probability for CH selection. The proposed probabilities
for CH selection in EDDEEC are given as follows:
pi=





















poptEi(r)
(1+m(a+mob)) ¯
E(r)for Nml nodes if Ei(r)> Tabsolute
popt(1+a)Ei(r)
(1+m(a+mob)) ¯
E(r)for Adv nodes if Ei(r)> Tabsolute
popt(1+b)Ei(r)
(1+m(a+mob)) ¯
E(r)for Sup nodes if Ei(r)> Tabsolute
cpopt(1+b)Ei(r)
(1+m(a+mob)) ¯
E(r)for Nml, Adv, Sup nodes if
Ei(r)≤Tabsolute
(5.6)
The value of absolute residual energy level, Tabsolute, is written as:
Tabsolute =zEo(5.7)
where, zϵ(0,1). If z= 0 then I have traditional EDEEC. In reality, advanced and
super nodes may have not been a CH in rounds r, it is also probable that some
of them become CH and same is the case with the normal nodes. So, exact value
14

of zis not sure. However, numerous simulations using random topologies, I try
to estimate the nearest value of zby varying it for best result based on first dead
node in the network and find best result for z= 0.7. Therefore, Tabsolute = (0.7)Eo.
In last, probability function in eq. 5.6 describes using absolute residual energy,
cis variable controlling the cluster number. If cis higher, then there are large
number of CH transmitting directly to the BS, thus network scenario becomes
same like the DC because all nodes will be CH. If c= 0, then there is no CH and
all nodes are transmitting directly to the BS same as in DC. Network performance
will decrease for both very high and very low value of c. To solve this I have done
numerous simulations to find best value of c and find it at c= 0.025, for better
network efficiency as shown in Fig. 5.1.
Figure 5.1: Round First Node dies when c is varying
15

Figure 5.2: Network Topology of EDDEEC
16

Chapter 6
BEENISH and i-BEENISH
Protocols
6.1 BEENISH Protocol
BEENISH implements the same concept as in DEEC, in terms of selecting CH
which depends on residual energy level of nodes with respect to average energy
of network. However, DEEC considers two types of nodes; normal and advanced
nodes. BEENISH uses the concept of four types of nodes; normal, advanced, super
and ultra-super nodes.
Let nishows the rounds for a node sito become CH, I refer it as rotating epoch.
CH has more energy consumption as compare cluster member nodes. In homo-
geneous networks, to ensure average poptNCHs in each round, LEACH let every
node si(i= 1,2, ....N ) to become CH once in every ni=1
popt rounds. During
operation of WSN all the nodes does not have the same remaining energy. So,
if the epoch niis kept equal for all nodes as in LEACH then energy is not well
distributed and nodes with low energy dies before high energy nodes. BEENISH
choose different epoch nifor different nodes according to their remaining energy
Ei(r). Nodes having high energy are more often elected as CH as compare to low
energy nodes. So, high energy nodes have smaller epoch nias compare to high en-
ergy nodes. In BEENISH ultra-super nodes are largely elected as CH as compare
to super, advanced and normal nodes. Super nodes are more often elected as CH
as compare to advanced and normal nodes. Advanced nodes have more election
probability to become CH than normal nodes. So, in this way energy consumed
by all nodes is equally distributed.
Let pi=1
niis probability of node for CH selection during epoch nirounds. When
17

all the nodes own same every level at each epoch, selecting the average probability
pito be popt ensure there are poptNCH each round and approximately all nodes
die at the same moment. If nodes are having different energy then nodes with
more energy have pilarger than popt.
In BEENISH, average energy of rth round can be obtained by Eq. 5.2 in previous
chapter, as supposed in DEEC:
To achieve the optimal number of CH in every round, node sidecides whether to
become a CH or not on basis of probability threshold calculated by Eq. 5.4 in
previous chapter,as supposed in [13, 19].
In real scenarios, WSNs have more greater than two or three energy levels of nodes.
In WSN due to random CH selection, large range of energy levels are created. So,
as much more energy levels I quantize and define different probability for every
energy level will lead to as much better results and lead to energy efficiency. In
BEENISH, I first time use concept of four level heterogeneous network having
normal, advanced, super and ultra-super nodes. The probabilities for four types
of nodes are given below:
pBEEN IS H
i=















poptEi(r)
(1+m(a+mo(−a+b+m1(−b+u)))) ¯
E(r)if siis the normal node
popt(1+a)Ei(r)
(1+m(a+mo(−a+b+m1(−b+u)))) ¯
E(r)if siis the advanced node
popt(1+b)Ei(r)
(1+m(a+mo(−a+b+m1(−b+u)))) ¯
E(r)if siis the super node
popt(1+u)Ei(r)
(1+m(a+mo(−a+b+m1(−b+u)))) ¯
E(r)if siis the ultra −super node
(6.1)
In the above expression I find that nodes with greater remaining energy Ei(r) at
round rare more possibly to become CH as compare to low energy nodes. The aim
of this mechanism is to efficiently divide the energy consumption in the network
and extend the stability period which is defined by first node die and network
lifetime defined by last node die from the start of WSN.
Simulation results proves BEENISH protocol to be the more efficient protocol than
DEEC, DDEEC and EDEEC for WSN containing four and multi level heterogene-
ity in terms of first node die and last node die. However, performance of BEENISH
can be more enhanced. It is possible on one moment that ultra-super, super and
advanced nodes have same energy as that of normal nodes at some round. Al-
though BEENISH continuous to penalize just the ultra-super, super and advanced
nodes respectively due to having higher probabilities. So, ultra-super nodes are
more penalized than super and advanced nodes. Same super nodes are more pe-
18

nalized than advanced nodes. This is not the optimal way, as high energy nodes
continuously elected as CH in every round will die quickly than others. So, this
case must be avoided and probabilities of ultra-super, super and advanced nodes
must be lowered when their energy become same as normal nodes.
6.2 i-BEENISH Protocol
The variance between DEEC, DDEEC, EDEEC and BEENISH is localized in Eq.
6.1 that defines probabilities to become CH for current round. Objective of this
expression is to divide energy consumption over network, efficiently increasing
stability period and lifetime of network. However, after some rounds, some ultra-
super, super and advanced nodes have same residual energy level as of normal
nodes due to CH selection. Although BEENISH continues to punish just ultra-
super, super and advanced nodes. This is not the optimal way, after some rounds
due to continues selection of ultra-super, super and advanced nodes as CH, they
die quickly as compare to normal nodes.
To avoid this unbalanced case I propose Improved BEENISH (i-BEENISH) with
some changing in probability function defined by BEENISH. These changes are
based on absolute residual energy Tabsolute, which is the value under which ultra-
super, super and advanced nodes have same energy level as that of normal nodes.
This idea specifies that under Tabsolute all ultra-super, super, advanced and normal
nodes have same probability for CH selection. Proposed probabilities of being CH
in BEENISH are given as below:
19

piBEEN IS H
i=

























































poptEi(r)
(1+m(a+mo(−a+b+m1(−b+u)))) ¯
E(r)for N ormal nodes
if Ei(r)> Tabsolute
popt(1+a)Ei(r)
(1+m(a+mo(−a+b+m1(−b+u)))) ¯
E(r)for Advanved nodes
if Ei(r)> Tabsolute
popt(1+b)Ei(r)
(1+m(a+mo(−a+b+m1(−b+u)))) ¯
E(r)for Super nodes
if Ei(r)> Tabsolute
popt(1+u)Ei(r)
(1+m(a+mo(−a+b+m1(−b+u)))) ¯
E(r)for U ltra −super nodes
if Ei(r)> Tabsolute
c×popt(1+u)Ei(r)
(1+m(a+mo(−a+b+m1(−b+u)))) ¯
E(r)for N ormal, Advanced, Super
and U ltra −super nodes
if Ei(r)≤Tabsolute
(6.2)
Value of absolute residual energy level, Tabsolute, is as follows in Eq. 6.3:
Tabsolute =zEo(6.3)
Where value of zlies between [0,1]. If I take z= 0 then I have traditional
BEENISH. In reality and through lot of simulations I observed that all the ultra-
super, super and advanced nodes have not been CH. Same is the case with normal
nodes, it is possible that some of them will become CH. So, exact value of zcannot
be confirmed. However, through numerous simulations using random topologies,
I try to find the nearest value of zby varying it for best results based on first
dead node in the network and find best result at z= 0.71. In Fig 6.1, using the
parameters described in Table 6.1, I show the first node dies variation in function
by varying zin numerous simulations.
Tabsolute = 0.71 ×Eo(6.4)
cis a real positive integer, controls the number of CH during each round. If cis
higher then method will be like Direct Communication (DC) because almost all
the nodes will be CH and they have to send data directly to BS. On the other
hand, if cis lower then method will be again like DC, there will be almost no CH
and all the nodes have to send data directly to BS. DC consumes lot of energy
so I avoid it. To solve this problem I have to find correct value of cwhich gives
best results in terms of first node die. Through numerous simulations, I find best
20

Figure 6.1: Round First Node dies when z is varying
Table 6.1: Simulation Parameters
Parameters Values
Network field 100 m,100 m
Number of nodes 100
Eo(initial energy of nor-
mal nodes)
0.5J
Message size 4000 bits
Eelec 50nJ/bit
Efs 10nJ/bit/m2
Eamp 0.0013pJ/bit/m4
EDA 5nJ/bit/signal
do(threshold distance) 70m
Popt 0.1
results in terms of first node die at c= 0.25 by varying cbetween range [0,1]. Fig.
6.2 shows how ceffect the round value of first node die.
21

Figure 6.2: Round First Node dies when c is varying
Figure 6.3: Network Topology of BEENISH and i-BEENISH
6.3 Performance Criteria Used
The performance parameters used for evaluation of clustering protocols for het-
erogeneous WSNs are stability period, network lifetime and packets sent to BS.
Stability period is period from start of network until death of first node whereas,
22

the instability period is period from death of first node till last node die.
Lifetime is a parameter which shows that all the nodes of each type in the WSN
have not yet consumed all of its energy.
Packets sent to BS or throughput is the measure that how many packets are
received by BS for each protocol in heterogeneous WSNs.
Assuming all nodes are stationary and of equal importance, number of nodes in
different protocols at different locations like at first, tenth and last position showed
difference in their values.
23

Chapter 7
Simulation Results
7.1 Simulation Results of EDDEEC
In this section, I present simulation result for DEEC, DDEEC, EDEEC and
EDDEEC for three level and multi level heterogeneous WSNs using MATLAB.
WSNs consist of N= 100 nodes which are placed randomly in area of dimension
100m×100m. Assuming all nodes are stationary and of equal importance, num-
ber of nodes in different protocols at different locations like at first, tenth and last
position shows difference in their values.
In heterogeneous WSNs, I have used radio parameters as mentioned in Table. 6.1
for different protocols deployed in WSNs and estimated performance for the case
of three level heterogeneous WSNs. I first observe performance of DEEC, DDEEC,
EDEEC and EDDEEC for three level heterogenous WSNs.
7.1.1 Case 1: m= 0.5, mo= 0.4, a = 1.5and b = 3
Now considering network which contains 50 normal nodes having Eoenergy, 30
advanced nodes having 1.5 times extra energy than normal nodes and 20 super
nodes containing 3 times extra energy than normal nodes. It is obvious as from
Fig. 7.1 and 7.2 that shows simulation results of alive and dead during lifetime of
the network. First node for DEEC, DDEEC, EDEEC and EDDEEC dies at 526,
1252, 1291 and 1510 rounds, respectively, and all nodes dies at 5380, 4523, 6920
and 6920 rounds, respectively. Fig. 7.3 shows that data sent to the BS is more
for EDDEEC then rest of the protocols. It is obvious from results that EDDEEC
is most efficient of all protocols in terms of stability period of network, Network
life time and Packets sent to the BS.
24

Figure 7.1: Nodes Alive During Network Lifetime
Figure 7.2: Nodes Dead During Network Lifetime
7.1.2 Case 2: m= 0.8, mo= 0.6, a = 2.0and b = 3.5
In case 2, I consider a network containing 20 normal nodes having Eoenergy, 32
advanced nodes having 2.0 times extra energy than normal nodes and 48 super
nodes containing 3.5 times extra energy than normal nodes. Fig. 7.4 and 7.5
depict number of alive and dead nodes during lifetime of network. First node for
DEEC, DDEEC, EDEEC and EDDEEC dies at 969, 1355, 1432 and 1717 rounds,
respectively, and all nodes dies at 5536, 5673, 8638 and 8638 rounds respectively.
Fig. 7.6 shows that data sent to BS is more for EDDEEC then rest of chosen
25

Figure 7.3: Packets Sent to Base Station
protocols. It is obvious from results that EDDEEC is most efficient all protocols
in terms of stability period of network, Network life time and Packets sent to BS
even in case of network containing more super and advanced nodes as compare to
normal nodes.
Figure 7.4: Nodes Alive During Network Lifetime
26

Figure 7.5: Nodes Dead During Network Lifetime
Figure 7.6: Packets Sent to Base Station
7.1.3 Case 3: m= 0.3, mo= 0.2, a = 1.2and b = 2.5
In this case, I place 70 normal nodes having Eoenergy, 24 advanced nodes having
1.2 times extra energy than normal nodes and 6 super nodes containing 2.5 times
extra energy than normal nodes. Fig. 7.7 and 7.8 show nodes alive and dead during
lifetime of network. First node for DEEC, DDEEC, EDEEC and EDDEEC dies at
1115, 1209, 1400 and 1682 rounds, respectively, and all nodes dies at 4693, 3726,
5798 and 5789 rounds, respectively. Fig. 7.9 shows that data sent to BS is more for
EDDEEC then rest of the protocols. It is obvious from results that EDDEEC is
27

most efficient among all protocols in terms of stability period of network, Network
life time and Packets sent to BS even in case of network containing less super and
advanced nodes, as compare to normal nodes.
Figure 7.7: Nodes Alive During Network Lifetime
Figure 7.8: Nodes Dead During Network Lifetime
7.1.4 Case 4: Multi-level Heterogenity
To consider multi-level heterogenity, I set nodes within set [0.5, 2]. Results given
in Fig. 7.10 and 7.11 shows alive and dead nodes during lifetime of network.
28

Figure 7.9: Packets Sent to Base Station
First node for DEEC, DDEEC, EDEEC and EDDEEC dies at 1184, 1307, 1353
and 1448 rounds, respectively, and all nodes dies at 3940, 3212, 4293 and 5210
rounds, respectively. Fig. 7.12 shows that data sent to BS is more for EDDEEC
then DEEC, DDEEC and EDEEC. It is obvious from results that EDDEEC is
most efficient of all protocols in terms of stability period of network, Network
life time and Packets sent to BS even in case of network containing multi-level
heterogeneity.
Figure 7.10: Nodes Alive During Network Lifetime
29

Figure 7.11: Nodes Dead During Network Lifetime
Figure 7.12: Packets Sent to Base Station
30

7.2 Simulation Results of BEENISH and i-BEENISH
This section evaluates the performance of BEENISH and i-BEENISH protocols
using MATLAB. I consider a WSN containing N= 100 nodes randomly dis-
tributed inside 100m×100mfield. For simplicity, I assume all nodes are either
fixed or micro-mobile and ignore energy loss due to signal collision and interference
between signals of different nodes that are due to dynamic random channel con-
ditions. My simulations use radio parameters mentioned in Table 6.1. Protocols
compared with BEENISH and i-BEENISH include DEEC, DDEEC and EDEEC.
I estimated performance for the case of four level and multi-level heterogeneous
WSNs. I first observe performance of DEEC, DDEEC, EDEEC and BEENISH for
four level and multi-level heterogenous WSNs. BEENISH is then compared with
i-BEENISH for four and multi-level heterogenous WSNs. In last i-BEENISH is
compared with DEEC, DDEEC and EDEEC for four and multi-level heterogenous
WSNs.
7.2.1 Case 5: m= 0.5, mo= 0.3, m1= 0.2, a = 1.5, b =
2.0and u = 2.5
Now considering the case 5, having network parameters m= 0.5, mo= 0.3, m1=
0.2, a = 1.5, b = 2.0and u = 2.5 containing 50 normal nodes having Eoenergy,
35 advanced nodes having 1.5 times extra energy than normal nodes, 12 super
nodes containing 2 times extra energy than normal nodes and 3 ultra-super nodes
containing 2.5 times extra energy than normal nodes. By simulation it is obvious
as from Fig. 7.13 and 7.14 showing alive and dead nodes during lifetime of the
network. First node for DEEC, DDEEC, EDEEC and BEENISH dies at 1358,
1489, 1381 and 1687 rounds, respectively. All nodes dies at 6530, 4788, 7455 and
7446 rounds, respectively. Fig. 7.15 shows BEENISH sends more data to BS than
DEEC, DDEEC and EDEEC. It is obvious from results that BEENISH is efficient
as compare to all protocols in terms of stability period of network, Network life
time and Packets sent to the BS.
7.2.2 Case 6: m= 0.6, mo= 0.5, m1= 0.3, a = 2.0, b =
2.5and u = 3.0
Now considering case 6, having network parameters m= 0.6, mo= 0.5, m1=
0.3, a = 2.0, b = 2.5and u = 3.0 containing 40 normal nodes having Eoenergy, 30
31

Figure 7.13: Nodes Dead During Network Lifetime
Figure 7.14: Nodes Alive During Network Lifetime
advanced nodes having 2 times extra energy than normal nodes, 21 super nodes
containing 2.5 times extra energy than normal nodes and 9 ultra-super nodes
containing 3 times extra energy than normal nodes. Fig. 7.16 and 7.17 showing
nodes alive and dead during lifetime of network. First node for DEEC, DDEEC,
EDEEC and BEENISH dies at 1295, 1565, 1344 and 2273 rounds, respectively.
All nodes dies at 7659, 5541, 8723 and 8709 rounds, respectively. Fig. 7.18 shows
BEENISH sends more data to BS than DEEC, DDEEC and EDEEC. It is obvious
from results that BEENISH is efficient of all protocols in terms of stability period
of network, Network life time and Packets sent to BS even in case of network
32

Figure 7.15: Packets Sent to BS
containing more super and advanced nodes as compare to normal nodes.
Figure 7.16: Nodes Dead During Network Lifetime
7.2.3 Case 7: Multi-level Heterogeneity
Now considering case 7, having multiple energy levels of nodes within set [1.5,2.5].
Fig. 7.19 and 7.20 showing nodes alive and dead during lifetime of network. First
node for DEEC, DDEEC, EDEEC and BEENISH dies at 1293, 1324, 1278 and
1464 rounds, respectively. All nodes dies at 5751, 4455, 6963 and 9623 rounds,
respectively. Fig. 7.21 shows BEENISH sends extra data to BS than DEEC,
33

Figure 7.17: Nodes Alive During Network Lifetime
Figure 7.18: Packets Sent to BS
DDEEC and EDEEC. It is obvious from results that BEENISH is efficient of all
protocols in terms of stability period of network, Network life time and Packets
sent to BS even in case of network containing multi-level heterogeneity.
34

Figure 7.19: Nodes Dead During Network Lifetime
Figure 7.20: Nodes Alive During Network Lifetime
7.2.4 Case 8: m= 0.5, mo= 0.3, m1= 0.2, a = 1.5, b =
2.0and u = 2.5
Now considering the case 8, having network parameters m= 0.5, mo= 0.3, m1=
0.2, a = 1.5, b = 2.0and u = 2.5 containing 50 normal nodes having Eoenergy,
35 advanced nodes having 1.5 times extra energy than normal nodes, 12 super
nodes containing 2 times extra energy than normal nodes and 3 ultra-super nodes
containing 2.5 times extra energy than normal nodes. By simulation it is obvious
as from Fig. 7.22 and 7.23 showing nodes alive and dead during lifetime of the
35

Figure 7.21: Packets Sent to BS
network. First node for BEENISH and i-BEENISH dies at 1269 and 1632 rounds,
respectively. All nodes dies at 7396 and 7396 rounds, respectively. Fig. 7.24 shows
i-BEENISH sends more data to BS than BEENISH. It is obvious from results that
i-BEENISH is more efficient as compare to BEENISH in terms of stability period
of network and Packets sent to the BS.
Figure 7.22: Nodes Dead During Network Lifetime
36

Figure 7.23: Nodes Alive During Network Lifetime
Figure 7.24: Packets Sent to BS
7.2.5 Case 9: m= 0.6, mo= 0.5, m1= 0.3, a = 2.0, b =
2.5and u = 3.0
Now considering case 9, having network parameters m= 0.6, mo= 0.5, m1=
0.3, a = 2.0, b = 2.5and u = 3.0 containing 40 normal nodes having Eoenergy, 30
advanced nodes having 2 times extra energy than normal nodes, 21 super nodes
containing 2.5 times extra energy than normal nodes and 9 ultra-super nodes
containing 3 times extra energy than normal nodes. Fig. 7.25 and 7.26 showing
37

nodes alive and dead during lifetime of network. First node for BEENISH and
i-BEENISH dies at 1278 and 1736 rounds, respectively. All nodes dies at 9637
and 9632 rounds, respectively. Fig. 7.27 shows i-BEENISH sends more data to
BS than BEENISH. It is obvious from results that i-BEENISH is more efficient
then BEENISH in terms of stability period of network and Packets sent to BS.
Figure 7.25: Nodes Dead During Network Lifetime
Figure 7.26: Nodes Alive During Network Lifetime
38

Figure 7.27: Packets Sent to BS
7.2.6 Case 10: Multi-level Heterogeneity
Now considering case 10, having multiple energy levels of nodes within set [1.5,2.5].
Fig. 7.28 and 7.29 showing nodes alive and dead during lifetime of network. First
node for BEENISH and i-BEENISH dies at 1429 and 1497 rounds, respectively.
All nodes dies at 7897 and 7908 rounds, respectively. Fig. 7.30 shows i-BEENISH
sends more data to BS than BEENISH. It is obvious from results that i-BEENISH
is most efficient of all protocols in terms of stability period of network, Network
life time and Packets sent to BS even in case of network containing multi-level
heterogeneity.
7.2.7 Case 11: m= 0.5, mo= 0.3, m1= 0.2, a = 1.5, b =
2.0and u = 2.5
Now considering the case 11, having network parameters m= 0.5, mo= 0.3, m1=
0.2, a = 1.5, b = 2.0and u = 2.5 containing 50 normal nodes having Eoenergy,
35 advanced nodes having 1.5 times extra energy than normal nodes, 12 super
nodes containing 2 times extra energy than normal nodes and 3 ultra-super nodes
containing 2.5 times extra energy than normal nodes. By simulation it is obvious
as from Fig. 7.31 and 7.32 showing nodes alive and dead during lifetime of the
network. First node for DEEC, DDEEC, EDEEC and i-BEENISH dies at 1336,
1614, 1408 and 2238 rounds, respectively. All nodes dies at 6048, 4470, 7368 and
7373 rounds, respectively. Fig. 7.33 shows i-BEENISH sends more data to BS
39

Figure 7.28: Nodes Dead During Network Lifetime
Figure 7.29: Nodes Alive During Network Lifetime
than DEEC, DDEEC and EDEEC. It is obvious from results that i-BEENISH is
most efficient as compare to all protocols in terms of stability period of network,
Network life time and Packets sent to the BS.
40

Figure 7.30: Packets Sent to BS
Figure 7.31: Nodes Dead During Network Lifetime
7.2.8 Case 12: m= 0.6, mo= 0.5, m1= 0.3, a = 2.0, b =
2.5and u = 3.0
Now considering case 12, having network parameters m= 0.6, mo= 0.5, m1=
0.3, a = 2.0, b = 2.5and u = 3.0 containing 40 normal nodes having Eoenergy, 30
advanced nodes having 2 times extra energy than normal nodes, 21 super nodes
containing 2.5 times extra energy than normal nodes and 9 ultra-super nodes
containing 3 times extra energy than normal nodes. Fig. 7.34 and 7.35 showing
nodes alive and dead during lifetime of network. First node for DEEC, DDEEC,
41

Figure 7.32: Nodes Alive During Network Lifetime
Figure 7.33: Packets Sent to BS
EDEEC and i-BEENISH dies at 1209, 1707, 1384 and 2378 rounds, respectively.
All nodes dies at 7294, 5508, 9071 and 9080 rounds, respectively. Fig. 7.36
shows i-BEENISH sends more data to BS than DEEC, DDEEC and EDEEC. It
is obvious from results that i-BEENISH is most efficient of all protocols in terms
of stability period of network, Network life time and Packets sent to BS even in
case of network containing more super and advanced nodes as compare to normal
nodes.
42

Figure 7.34: Nodes Dead During Network Lifetime
Figure 7.35: Nodes Alive During Network Lifetime
7.2.9 Case 13: Multi-level Heterogeneity
Now considering case 13, having multiple energy levels of nodes within set [1.5,2.5].
Fig. 7.37 and 7.38 showing nodes alive and dead during lifetime of network. First
node for DEEC, DDEEC, EDEEC and i-BEENISH dies at 1212, 1550, 1423 and
1645 rounds respectively. All nodes dies at 5649, 4463, 7089 and 7091 rounds
respectively. Fig. 7.39 shows i-BEENISH sends more data to BS than DEEC,
DDEEC and EDEEC. It is obvious from results that i-BEENISH is efficient of all
protocols in terms of stability period of network, Network life time and Packets
43

Figure 7.36: Packets Sent to BS
sent to BS even in case of network containing multi-level heterogeneity.
Figure 7.37: Nodes Dead During Network Lifetime
Table 7.1: Comparing Performance of DEEC, DDEEC, EDEEC and EDDEEC for
m= 0.5, mo= 0.4, a = 1.5and b = 3
Protocols First node dies All nodes die Packet sent to BS
DEEC 526 5380 136468
DDEEC 1252 4523 108024
EDEEC 1291 6920 288186
EDDEEC 1510 6920 368405
44

Figure 7.38: Nodes Alive During Network Lifetime
Figure 7.39: Packets Sent to BS
Table 7.2: Comparing Performance of DEEC, DDEEC, EDEEC and EDDEEC for
m= 0.8, mo= 0.6, a = 2.0and b = 3.5
Protocols First node dies All nodes die Packet sent to BS
DEEC 969 5536 153936
DDEEC 1355 5673 141696
EDEEC 1432 8638 407642
EDDEEC 1717 8638 448084
45

Table 7.3: Comparing Performance of DEEC, DDEEC, EDEEC and EDDEEC for
m= 0.3, mo= 0.2, a = 1.2and b = 2.5
Protocols First node dies All nodes die Packet sent to BS
DEEC 1115 4693 108786
DDEEC 1209 3726 83179
EDEEC 1400 5798 233123
EDDEEC 1682 5798 316819
Table 7.4: Comparing Performance of DEEC, DDEEC, EDEEC and EDDEEC for
Multi-level Heterogenity
Protocols First node dies All nodes die Packet sent to BS
DEEC 1184 3940 110147
DDEEC 1307 3212 81515
EDEEC 1353 4293 228420
EDDEEC 1448 5210 313529
Table 7.5: Comparing Performance of DEEC, DDEEC, EDEEC BEENISH and i-
BEENISH for m= 0.5, mo= 0.3, m1= 0.2, a = 1.5, b = 2.0and u = 2.5
Protocols First node dies All nodes die Packet sent to BS
DEEC 1358 6530 256504
DDEEC 1489 4788 135973
EDEEC 1381 7455 372975
BEENISH 1687 7446 445377
i-BEENISH 2238 7373 489879
Table 7.6: Comparing Performance of DEEC, DDEEC, EDEEC BEENISH and i-
BEENISH for m= 0.6, mo= 0.5, m1= 0.3, a = 2.0, b = 2.5and u = 3.0
Protocols First node dies All nodes die Packet sent to BS
DEEC 1295 7659 339096
DDEEC 1565 5541 164979
EDEEC 1344 8723 449127
BEENISH 2273 8709 542399
i-BEENISH 2378 9260 551790
Table 7.7: Comparing Performance of DEEC, DDEEC, EDEEC BEENISH and i-
BEENISH for Multi-level Heterogenity
Protocols First node dies All nodes die Packet sent to BS
DEEC 1293 5751 191263
DDEEC 1324 4455 106309
EDEEC 1278 6963 300121
BEENISH 1464 6923 343650
i-BEENISH 1645 6982 402243
46

Chapter 8
Conclusion
In this thesis, EDDEEC protocol is proposed for WSNs. EDDEEC is adaptive
energy aware protocol which dynamically changes the probabilities of nodes to
become a CH in a balanced and efficient way to distribute equal amount of energy
between sensor nodes. Modifications largely improve performance of EDDEEC
as compared to DEEC, DDEEC and EDEEC in terms of stability period and life
time. I described BEENISH is a energy aware clustering based protocol designed
heterogenous WSNs. In BEENISH uses the concept of four types of nodes, elects
CH based of residual and average energy of the network. So, in BEENISH nodes
with high energy are largely selected as CH than low energy nodes. i-BEENISH
dynamically changes the CH election probabilities of high energy nodes when their
energy decreases sensibly. I perform extensive simulations to check the efficiency of
newly proposed protocols. The selected performance metrics for this analysis are
stability period, network lifetime and packets sent to BS. The simulation analysis
showed batter results which differentiate EDDEEC more efficient and reliable than
DEEC, DDEEC and EDEEC in three level heterogenous WSN. BEENISH and i-
BEENISH are proved to be more efficient than DEEC, DDEEC and EDEEC in
four level heterogenous WSN. i-BEENISH has longer stability period as compare
to BEENISH also.
47

References
[1] Qureshi. T. N, Javaid. N, Malik. M, Qasim. U, Khan. Z. A, On Perfor-
mance Evaluation of Variants of DEEC in WSNs, 7th International Confer-
ence on Broadband and Wireless Computing, Communication and Applica-
tions (BWCCA-2012), Victoria, Canada, 2012.
[2] Tauseef Shah, Nadeem Javaid, Talha Naeem Qureshi, “Energy Efficient Sleep
Awake Aware (EESAA) Intelligent Sensor Network Routing Protocol”, 15th
IEEE International Multi Topic Conference (INMIC’12), 2012, Pakistan.
[3] Aslam. M, Javaid. N, Rahim. A, Nazir. U, Bibi. A, Khan. Z. A, “Survey of
Extended LEACH-Based Clustering Routing Protocols for Wireless Sensor
Networks”, 5th International Symposium on Advances of High Performance
Computing and Networking (AHPCN-2012)in conjunction with 14th IEEE In-
ternational Conference on High Performance Computing and Communications
(HPCC-2012), 25-27 June, Liverpool, UK, 2012.
[4] I.F. Akyildiz, W. Su, Y. Sankarasubramaniam, E. Cayirci, “Wireless sensor
networks: a survey, Computer Networks”, 38 (4) (2002) 393-422.
[5] Aslam. M, Shah. T, Javaid. N, Rahim. A, Rahman. Z, Khan. Z. A, “CEEC:
Centralized Energy Efficient Clustering A New Routing Protocol for WSNs”,
Poster Session of 9th Annual IEEE Communications Society Conference on
Sensor, Mesh and Ad Hoc Communications and Networks (SECON2012),
Seoul, Korea, 2012.
[6] Latif. K, Jaffar. M, Javaid. N, Saqib. M. N, Qasim. U, Khan. Z. A, “Perfor-
mance Analysis of Hierar- chical Routing Protocols in Wireless Sensor Net-
works”, 5th International Workshop on Next Generation of Wireless and Mo-
bile Networks (NGWMN 2012) in conjunction with 7th IEEE International
Conference on Broadband and Wireless Computing, Communication and Ap-
plications (BWCCA 2012), Victoria, Canada, 2012.
48

[7] H. Chan, A. Perrig, ACE: “An Emergent Algorithm for Highly Uniform Cluster
Formation”, Proceedings of the First European Workshop Sensor Networks
(EWSN), January 2004.
[8] M. Younis, M. Youssef, K. Arisha, “Energy-aware management in cluster-based
sensor networks”, Computer Networks 43 (5) (2003), 649668.
[9] P. Krishna, N.H. Vaidya, M. Chatterjee, D. Pradhan, “A cluster-based ap-
proach for routing in dynamic networks, ACM SIGCOMM Computer Com-
munication Review”, 27 (2) (1997) 49-65.
[10] B. McDonald, T. Znati, “Design and performance of a distributed dynamic
clustering algorithm for Ad-Hoc networks”, in: Proceedings of the Annual
Simulation Symposium, 2001.
[11] V. Mhatre, C. Rosenberg, D. Kofman, R. Mazumdar, N. Shroff, “Design of
surveillance sensor grids with a lifetime constraint”, in: 1st European Work-
shop on Wireless Sensor Networks (EWSN), Berlin, January 2004.
[12] W.R. Heinzelman, A.P. Chandrakasan, H. Balakrishnan, “Energy efficient
communication protocol for wireless micro sensor networks”, in: Proceedings
of the 33rd Hawaii International Conference on System Sciences (HICSS-33),
January 2000.
[13] W.R. Heinzelman, A.P. Chandrakasan, H. Balakrishnan, “An application-
specific protocol architecture for wireless microsensor networks”, IEEE Trans-
actions on Wireless Communications 1 (4) (2002) 660-670.
[14] S. Lindsey, C.S. Raghavenda, PEGASIS: “power efficient gathering in sensor
information systems”, in: Proceeding of the IEEE Aerospace Conference, Big
Sky, Montana, March 2002.
[15] O. Younis, S. Fahmy, HEED: “A hybrid, energy-efficient, distributed clus-
tering approach for ad hoc sensor networks”, IEEE Transactions on Mobile
Computing 3 (4) (2004) 660-669.
[16] G. Smaragdakis, I. Matta, A. Bestavros,“SEP: A Stable Election Protocol
for clustered heterogeneous wireless sensor network”, in: Second International
Workshop on Sensor and Actor Network Protocols and Applications (SANPA
2004), 2004.
[17] Khan. A. A, Javaid. N, Qasim. U, Lu. Z, Khan. Z. A, “HSEP: Heterogeneity-
aware Hierarchical Stable Election Protocol for WSNs”, 3rd International
Workshop on Advances in Sensor Technologies, Systems and Applica-
tions (ASTSA-2012) in conjunction with 7th IEEE International Confer-
49

ence on Broadband and ireless Computing, Communication and Applications
(BWCCA 2012), Victoria, Canada, 2012.
[18] Aasia Kashaf, Nadeem Javaid, Zahoor Ali Khan and Imran Ali Khan, “TSEP:
Threshold-sensitive Stable Election Protocol for WSNs”, 10th IEEE Interna-
tional Conference on Frontiers of Information Technology (FIT 12), 2012, Pak-
istan.
[19] L. Qing, Q. Zhu, M. Wang, “Design of a distributed energy-efficient clustering
algorithm for heterogeneous wireless sensor network”, ELSEVIER, Computer
Communications 29, 2006, pp 2230- 2237.
[20] Elbhiri, B. , Saadane, R. , El Fkihi, S. , Aboutajdine, D. “Developed Dis-
tributed Energy-Efficient Clustering (DDEEC) for heterogeneous wireless sen-
sor networks”, in: 5th International Symposium on I/V Communications and
Mobile Network (ISVC), 2010.
[21] Parul Saini, Ajay.K.Sharma, “E-DEEC- Enhanced Distributed Energy Effi-
cient Clustering Scheme for heterogeneous WSN”, in: 2010 1st International
Conference on Parallel, Distributed and Grid Computing (PDGC - 2010).
50