On Sink Mobility Trajectory in Clustering Routing ... · xof Higher Education Comission (HEC) rules...
Transcript of On Sink Mobility Trajectory in Clustering Routing ... · xof Higher Education Comission (HEC) rules...
i
On Sink Mobility Trajectory in Clustering Routing Protocols in WSNs
By Miss Qurat Ul Ain
Registration Number: CIIT/FA11-REE-058/ISB MS Thesis
In Electrical Engineering
COMSATS Institute of Information Technology Islamabad – Pakistan
FALL, 2012
ii
On Sink Mobility Trajectory in Clustering Routing
Protocols in WSNs
A Thesis presented to COMSATS Institute of Information Technology
In partial fulfillment of the requirement for the degree of
MS (Electrical Engineering)
By
Ms. Qurat Ul Ain
CIIT/FA11-REE-058/ISB
Fall, 2012
iii
On Sink Mobility Trajectory in Clustering Protocols in WSNs
A Graduate Thesis submitted to Department of Electrical Engineering as partial fulfillment of the requirement for the award of M. S. Degree
(Electrical Engineering).
Name Registration Number Miss Qurat Ul Ain CIIT/FA11-REE-058/ISB
Co-supervisor: Dr. Nadeem Javaid, Assistant Professor,
Center for Advanced Studies in Telecommunications (CAST), COMSATS Institute of Information Technology (CIIT),
Islamabad Campus, December, 2012
Supervisor: Dr. Mahmood Ashraf Khan,
Director, Center for Advanced Studies in Telecommunications (CAST),
COMSATS Institute of Information Technology (CIIT), Islamabad Campus,
December, 2012
iv
Final Approval
This thesis entitled
On Sink Mobility Trajectory in Clustering Routing Protocols in WSNs
By Miss Qurat Ul Ain
CIIT/FA11-REE-058/ISB
has been approved for the COMSATS Institute of Information Technology, Islamabad
External Examiner: __________________________________ (To be decided)
Co-Supervisor: ________________________ Dr. Nadeem Javaid /Assistant professor, Center for Advanced Studies in Telecommunications (CAST), CIIT, Islamabad.
Supervisor: ________________________ Dr. Mahmood Ashraf Khan/Director, 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 Miss Qurat Ul Ain, CIIT/FA11-REE-058/ISB herebyxdeclare that I havexproduced the workxpresented 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 Higher Education Comission (HEC) rulesxon research hasxoccurred in thisxthesis, I shall be liablexto punishablexaction under the plagiarismxrules of the HEC.
Date: ________________ ________________ Miss Qurat Ul Ain CIIT/FA11-REE-058/ISB
vi
Certificate
It is certified that Miss Qurat Ul Ain, CIIT/FA11-REE-058/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: _________________ Co-Supervisor:____________________ Dr. Nadeem Javaid /Assistant professor, Center for Advanced Studies in Telecommunications (CAST), CIIT, Islamabad.
Supervisor: ________________________ Dr. Mahmood Ashraf Khan/Director, 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 parents.
viii
ACKNOWLEDGMENT I am heartily grateful to my supervisor, Dr. Mahmood Ashraf Khan, and co-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.
Miss Qurat Ul Ain CIIT/FA11-REE-058/ISB
ix
List of Publications
[1] Ain.Q, Ikram. A, Javaid. N, Qasim. U, Khan. Z. A, “Modeling Propagation Characteristics for Arm- Motion in Wireless Body Area Sensor Networks”, submitted in 7th International Conference on Broadband and Wireless Computing, Communication and Applications (BWCCA-2012), Victoria, Canada, 2012.
[2] Ain.Q, Javaid.N, "On Sink Mobility Trajectory in Clustering Routing Protocols in WSNs", submitted in, 10th IEEE International Conference on Wireless On-demand Network Systems and Services (WONS'13), March 18-20, 2013, Banff, Canada.
x
List of Abbreviations
WSNs Wireless Sensor Networks
LEACH Low Energy Adaptive Clustering Hierarchy
TEEN Threshold Sensitive Energy Efficient Routing Protocol
SEP Stable Election Probability
DEEC Distributed Energy Efficient Clustering
CAMPTEEN Clustering and Multi-Hop Protocol in Threshold Sensitive Energy Efficient Sensor Network
HTEEN Hierarchical Threshold Sensitive Energy Efficient Sensor Network
IR Infra Red
CH Cluster Head
BS Base Station
Far-Zone LEACH FZ-LEACH
MRP Minimum Reachabilty Power
AMRP Average Minimum Reachabilty Power
xi
Abstract
Energy efficient routing protocols are apparently used in Wireless Sensor networks (WSNs). The area of WSNs is one of the emerging and fast growing fields which brought sensor nodes with multi functions. In this work, we examine some protocols related to homogeneous and heterogeneous networks. I compare five clustering routing protocols; Low Energy Adaptive Clustering Hierarchy (LEACH), Threshold Sensitive Energy Efficient Sensor Network (TEEN), Distributed Energy Efficient Clustering (DEEC) and two variants of TEEN which are Clustering and Multi-Hop Protocol in Threshold Sensitive Energy Efficient Sensor Network (CAMP-TEEN) and Hierarchical Threshold Sensitive Energy Efficient Sensor Network (H-TEEN) to evaluate the efficiency of different clustering schemes. Concept of mobile sink is introduced to enhance the network life time and energy efficiency of these routing protocols. Two scenarios are discussed to compare the performances of routing protocols; in first scenario static sink is implanted and mobile sink is used in the second scenario. I perform analytical simulations in MATLAB by using different performance metrics such as number of alive nodes, number of dead nodes and throughput.
Contents
1 Introduction 1
2 Background and Motivation for Thesis 3
2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3 Overview of Routing Protocols for WSNs 8
3.1 Energy Efficient Routing Protocols . . . . . . . . . . . . . . . . . . 8
3.1.1 LEACH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.1.2 TEEN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.1.3 DEEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1.4 H-TEEN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.1.5 CAMP-TEEN . . . . . . . . . . . . . . . . . . . . . . . . . 14
4 Sink Mobility 16
4.1 Data Aggregation Process in Sink Mobility . . . . . . . . . . . . . . 16
4.2 Strategies of Sink Mobility . . . . . . . . . . . . . . . . . . . . . . . 17
4.3 Model for Sink Mobility . . . . . . . . . . . . . . . . . . . . . . . . 17
4.3.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.3.2 System Model . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.3.3 Persuade Problem . . . . . . . . . . . . . . . . . . . . . . . . 19
4.3.4 Maximization of Network Lifetime . . . . . . . . . . . . . . . 21
4.3.4.1 Network Lifetime of Static Sink Model . . . . . . . 21
4.3.4.2 Network Lifetime of Mobile Sink Model . . . . . . 22
4.3.4.3 Maximize Lifetime in Threshold based Protocol . . 23
4.3.4.4 Maximize Lifetime for scalable Model . . . . . . . . 24
5 Simulations and Results 26
5.1 Radio Dissipation Model . . . . . . . . . . . . . . . . . . . . . . . . 26
5.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.2.1 Analyzing Scalability . . . . . . . . . . . . . . . . . . . . . . 33
xii
5.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
References 39
xiii
List of Figures
4.1 Sink Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.1 Radio Energy Dissipation Model . . . . . . . . . . . . . . . . . . . . 27
5.2 Heterogeneous Network . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.3 Homogeneous Network . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.4 Number of alive nodes with static sink (100 nodes and 100m × 100m ) 30
5.5 Number of dead nodes with static sink (100 nodes and 100m × 100m ) 30
5.6 Throughput of protocols with static sink (100 nodes and 100m ×100m ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.7 Number of alive nodes with mobile sink (100 nodes and 100m ×100m ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.8 Number of dead nodes with mobile sink (100 nodes and 100m ×100m ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.9 Throughput of protocols with mobile sink (100 nodes and 100m ×100m ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.10 Number of alive nodes with static sink (200 nodes and 100m × 100m ) 34
5.11 Number of dead nodes with static sink (200 nodes and 100m × 100m ) 35
5.12 Throughput of protocols with static sink (200 nodes and 100m ×100m ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5.13 Number of alive nodes with mobile sink (200 nodes and 100m ×100m ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.14 Number of dead nodes with mobile sink (200 nodes and 100m ×100m ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.15 Throughput of protocols with mobile sink (200 nodes and 100m ×100m ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.16 Throughput of all protocols showing in bar graph . . . . . . . . . . 37
5.17 Number of alive nodes showing in bar graph . . . . . . . . . . . . . 38
xiv
List of Tables
3.1 Features of DEEC, LEACH, TEEN and its Variants . . . . . . . . . 15
5.1 Values of Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.2 Comparing Performance of DEEC, LEACH, TEEN and its Variants 38
xv
Chapter 1
Introduction
WSNs have become prominent now a days because they hold the potential to re-
form our life. The working of sensor networks have many issues to handle including
networking, information management and signal processing. These networks are
deployed in such environments which are resource constrained such as battery
limitations. Advancement in wireless networking yields suitable environment for
commercial and military applications. In WSNs, nodes are equipped with micro-
processor and sensing devices for different applications such as acoustic, seismic,
Infra Red (IR) and sensing devices. All nodes communicate with neighboring
nodes within its range. Sensors can aggregate information from physical environ-
ment such as temperature, vibration, humidity and other changes. This network is
more expensive than other conventional networks because it generates new types
of traffic. The difference between sensor networks and Ad -hoc networks are; num-
ber of sensor nodes are higher than Ad-hoc, topology of network changes regularly,
senors are power limited and have limited hardware.
WSNs are composed of sensor nodes which can sense, compute, store and transceive
data of interest in environment. These sensor nodes are scattered in sensor field
and capable of transmitting data to sink. In many applications, WSNs are used
such as in military, environmental disaster areas. Some applications of WSNs are
temperature monitoring, landslide detection, air pollution and structural monitor-
ing. Two applications are discussed in this thesis such as the parameter used in
TEEN protocol is temperature and an extension of TEEN is designed for landslide
detection.
Clusters are formed by grouping nodes to increase the network lifetime and en-
ergy efficiency. Cluster formation provides two level hierarchy in which nodes are
present in first level and CHs are in second level. Sensor nodes transmit data
1
to CH by time to time. There are two parameters for CH selection algorithm:
probabilistic clustering algorithm and non-probabilistic algorithm. In probabilis-
tic algorithm, a probability is assigned to each node initially and they decide
individually to elect CH. In non-probabilistic algorithm, each node depends on its
neighboring node. Node receives signal from its neighboring nodes and evaluate
the strength. The node with high signal strength is eligible to be CH.
Nodes which are far away from sink consume more energy. Clustering technique
is one of the popular mechanisms in which nodes select a Cluster Head (CH) for
transmission. All nodes transmit their data to CH, where, it aggregates data
and send to the Base Station (BS). Less energy is consumed in clustering process
because all nodes are not used in transmission at large distance. The main idea of
clustering is to reduce the network traffic from node to sink. Clustering protocols
exhibit better performance than flat network topologies when we compare them
for energy consumption. In homogeneous networks, nodes have same energy level
and in heterogeneous networks, nodes have different energy levels. LEACH [1]
and TEEN [2] are the examples of homogeneous protocols, while SEP [3] (Stable
Election Probability) and DEEC [4] are heterogeneous Protocols. All clustering
techniques consist of two steps; steady state phase and setup phase. Clusters
are formed and CHs are elected in setup phase and in steady state phase, data
transmission process is done from CH to BS.
In earlier research, static sink is used to gather data in WSNs. The sensors near
to sink consume more energy and die earlier. The sink disconnects from network
but rest of sensors have sufficient residual energy which is wasted. To reduce this
problem, mobile sink is introduced which can gather data from whole network.
There are two types of sink mobility; uncontrollable and controllable mobility.
Controllable mobility is achieved by adding mobile intentionally in the network to
carry sink node. Uncontrollable mobility is by attaching the sink node in network
which is out of control.
In this work, I compare five routing protocols in terms of their stability period,
network life time and throughput. In case of stability period, DEEC performs
better than other protocols because it is a heterogeneous protocol which has nodes
with some extra energy known as advanced and super nodes. These nodes die later
than normal nodes so can transmit more packets. When we consider network life
time, H-TEEN [5] performs better than other protocols because, it is a threshold
based protocol where nodes only transmit when they sense this value and consume
less energy. I introduce the sink mobility in these protocols to prolong the lifetime
of network and compare performance of all protocols.
2
Chapter 2
Background and Motivation for
Thesis
2.1 Background
Vivek Ketyar proposed a model which is an improvement in original LEACH.
LEACH is a hierarchial clustering protocol in which clusters exist (very large and
very small in size) affects the performance of protocol. To overcome this prob-
lem a Far-Zone LEACH (FZ-LEACH) is introduced. Far-Zone consists of a group
of nodes which is deployed in sensor network where the energies are less than
the threshold values. A cluster based zone is formed on the basis of Minimum
Reachability Power (MRP) is cost of communication from member nodes to BS.
FAR-ZONE is considered when MRP is greater than Average Minimum Reacha-
bility Power (AMRP) . After the formation of FAR-ZONE, nodes select the Zone
head and transmit data. Zone head collects the information and forwards it to
BS. It improves the network life time and energy consumption. It saves energy by
30 percent than original LEACH [7].
Arezoo Yektaparast proposed an algorithm to improve the LEACH protocol. In
CELL-LEACH, clusters are further divided into subsection called cells. Each cell
has a cell head which is selected randomly as in LEACH. Cell head changes after
some rounds to consume less energy. Sensors in cell communicate only with cell
head and seven cell heads transmit data to CH. In this way sensors prevent from
communicating at large distance. In CELL-LEACH, there are some changes in
threshold value for CH selection criteria from original LEACH [8].
Mohammad Javad Hajikhani developed an algorithm for CH selection in WSNs.
In LEACH, each node makes decision to be CH independently but distribution of
3
even energy load is not probable. In this protocol, residual energy of each node
and its neighboring nodes is considered to make even energy. In this way, all nodes
should have information of neighboring nodes. Each node has different remaining
energy, this parameter is used to have infomation about node and its neighboring
nodes. Every node monitors variation of energy consumption between current
round and previous round and then decided to be CH. If the energy dissipation
of node in current round is greater than previous round, node can be CH for next
round. This model makes the network with even energy distribution and improves
the network life time [9].
ShenLin introduced a protocol which is based on the network having location in-
formation of all nodes to reduce the routing cost. When sensor nodes are deployed
in a network, it needs to get location information of all nodes through Global Po-
sitioning System (GPS) technology and send to BS. In LEACH, there are random
localization of clusters which can cause energy loss [10].
Bilal Abu Bakr developed LEACH-SM protocol which reduces the energy con-
sumption. Research shows that by adding spare nodes in network could enhance
the network lifetime. LEACH-SM is the spare management protocol which en-
hance LEACH by efficient management of spares and reduces energy consump-
tion. In spare election phase, each node decides to become an active primary node
or passive spare node. Spare nodes go asleep in WSN, they awake when primary
node consumes its energy to a predefined value. It reduces the duration of nodes
to be in active mode. There is less energy consumption because some nodes are
in sleep mode for a given time which extends lifetime [11].
Arati Manjeshwar proposed a model which is an extension of TEEN protocol
called APTEEN. TEEN protocol is used in only time critical applications but
APTEEN is used in both time critical and periodic data sending applications.
APTEEN have best features of both reactive and proactive networks. Once the
CHs are decided, they broadcast the attributes (parameters which nodes have to
sense), soft threshold and hard threshold values,count time (time period between
two successive reports sent by nodes and it accounts for proactive component) and
TDMA schedule in which nodes have to send data.[12].
Developed Distributed Energy Efficient Clustering (DDEEC) is developed by
Brahim Elbhiri uses same method for estimation of average energy in the network
and CH selection criteria based on residual energy as used in DEEC. The main
difference between DDEEC and DEEC is probability defined for normal nodes and
advance nodes. We find that nodes with high residual energy at current round
r have more chances to become CH, so, in this way nodes having higher energy
4
values or advanced nodes will become CH more often as compared to the nodes
with lower energy or normal nodes. A point comes in a network where advanced
nodes and normal nodes have same residual energy. DEEC will punish the ad-
vanced nodes after this point which is not an optimal way of energy distribution
because by doing so, advanced nodes are continuously elect as CH and they die
more quickly than normal nodes [13].
Paraul Saini enhanced the DEEC protocol by adding super nodes in it which is
named as Enhanced Distributed Energy Efficient Clustering (EDEEC). It extends
DEEC to three level heterogeneity by introducing super nodes. CH selection
criteria is based on three types of nodes super nodes, advanced nodes and normal
nodes. Super nodes have highest energy than normal and advanced nodes. These
nodes have high probability to be CH than normal and advanced nodes. EDEEC
performs better in terms of stability as compared to Stable Election Protocol
(SEP). It also extend the network lifetime [14].
Paraul Saini presented the protocol which is an extension of DEEC. Threshold
Distributed Energy Efficient Clustering (TDEEC) introduces the threshold value
of a node on which node decides to become CH. The threshold value is based on
average energy of network and remaining energy of node. This protocol increases
the stability period and network lifetime [15].
Z.Maria Wang explored the idea of sink mobility with energy constrained nodes
to increase the network lifetime. Paper presents the linear programming formu-
lation which determine the sink mobility and sojourn time at different locations
of network. It resolves the problem of energy depletion due to which nodes die
earlier and shortens the network lifetime [16].
Heinzelman et al. proposed a model in which he examines the movement of sink
from its first location to next location. This model minimizes the data loss and
makes the network fault tolerant. Heuristic for sojourn time is introduced to find
the performance of proposed algorithm [17].
Ioannis Chatzigiannakis developed the sink mobility in four different patterns.
Sink is moving to collect data from whole network. different patterns of mobility
are; partial random walk with limited multi-hop data propagation, random walk
and passive data collection, deterministic walk with multi-hop data propagation
and biased random walk with passive data collection . It leads to more efficient
hybrid solutions [18].
N. Vlajic proposed a model for load balancing in WSN by introducing sink mobil-
ity. Real-world networks including Zig-Bee are not zero-overhead network so, by
5
introducing sink mobility congestion is the main problem which occurs. Author
introduced simple mechanism which reduces the overhead related to mobility [19].
Aristotelis Giannakos proposed an algorithm to collect data from sensors by mobile
sink. When sink passes through nodes, it sends request to node for transmission.
In this way number of transmission become low. Another routing algorithm is
proposed named as controlled random walk and this model can be generalizes for
two or more sinks [20].
2.2 Motivation
Research has shown that nodes near the sink deplete their battery power faster
than the nodes apart due to heavy overhead of messages. Sensors nearby sink are
shared by more sensor to sink paths therefore consume more energy. This uneven
energy depletion causes energy holes and leads to shorten network lifetime. Nu-
merous research has been conducted to mitigate this problem such as MobiRoute
[24] routing protocol and Delay Tolerant Mobile Sink Model (DT-MSM) [25]. Sink
is varied at different locations of the network field area from center to border axis
in many routing schemes. Mobile sink is used in all models to improve throughput
and reduces energy consumption.
In LEACH, CH consumes more energy having large number of nodes than CH with
smaller number of nodes. To mitigate this issue MLEACH protocol is proposed, it
allows mobility of both sink and nodes. In this protocol, CH selection algorithm
is based on remaining energy of nodes. It improves the throughput and network
lifetime [21].
DEEC is a heterogeneous protocol having nodes with extra energy which can send
more packets to BS than other protocols. DEEC is compared with its variants to
evaluate the performances of all proposed protocols [22].
MSWSN model used mobile sink to gather data from static nodes. Model proposed
to move sink with relative distance, direction and speed. It increases the delivery
ratio, residual energy and network lifetime by one hop communication. Mobile sink
transmits Buffer space to each sensor node for storing data. WCDMA technology
is used to ensure energy efficient communication. Proposed model provides the
relative random motion in sensor network and avoid threats in the multi hop
communication [23].
MobiRoute routing protocol proposed a model with a predictable mobility of sink
to enhance the packet deliver ratio. The sink is located at any point of network
6
and the pause time of sink is longer than movement time. It have enough time to
gather the data from near by nodes. In this protocol all nodes should know the
topological changes created due to sink mobility [24].
In DT-MSM model, nodes can reschedule the data transmission until the sink
reaches at most appropriate position. This model implements where the nodes
can tolerate the delay in data delivery. As the sink location increases, optimal
network life increases but there is delay in receiving data [25].
A. Chakrabarti et al. proposed a model for power saving in WSNs using predicable
path of sink. The model is performed for single hop communication. It improves
the data gathering, and enhance the power consumption of network by using
queuing model [26].
This thesis discusses a frame work to maximize the network lifetime without delay
in transmission of data packets by using a mobile sink. The proposed model is
used in both single hop and multi hop communication. We consider both homo-
geneous and heterogeneous networks in which energy parameters are enhanced by
implementing the concept of sink mobility in routing protocols such as, LEACH,
TEEN, DEEC and variants of TEEN. Their performances are observed in MAT-
LAB. The position of sink is varied at different locations of the network field area
from center to border axis in many routing schemes. Proposed protocol imple-
ments sink on the top axis of the network region and compare the results of all
routing protocols.
7
Chapter 3
Overview of Routing Protocols
for WSNs
Clustered sensor networks are categorized in two types of network; homogeneous
and heterogeneous networks. In homogeneous network, all nodes are equipped
with same energy, there is no predetermined cluster heads that control the cluster.
Although this may be costly because each node is designed to be CH but good
due to its self independence of nominating the CHs. It is evident that in long
transmission of CH to sink, if the clustered nodes are same in all rounds, they will
be over-loaded with data and will expire before other nodes. One way to ensure
this is to rotate the CH randomly as proposed in LEACH. In this way, chance of
being CH is provided to all nodes and they will die at same time.
In heterogeneous network, there are many types of nodes other than normal nodes
such as advanced and super with different energy level and functionality. The use
of different energy is that it reduces the hardware cost of other network. Some
extra energy is given to nodes other than normal nodes so, they have more chances
to be CH than normal nodes. CHs are predetermined in heterogeneous protocol
that control the cluster.
3.1 Energy Efficient Routing Protocols
Ideally, sensor networks should perform its functionality as long as possible. Con-
ventional routing schemes are not feasible in terms of energy constrained. Energy
efficient scheme is used to improve the routing protocols. There are some energy
efficient routing protocols which are discussed below related to homogeneous and
heterogeneous networks.
8
3.1.1 LEACH
LEACH is a first clustering based protocol which provides same initial energy
among all nodes. After cluster formation, nodes selects one CH which require
minimum communication energy. LEACH uses random election of CH because if
CHs are fixed throughout the system they will die quickly and network discon-
nects. It aggregates the data from member nodes for transmitting to BS which
decreases the energy consumption and enhance lifetime of network. LEACH can
achieve factor of 8 in decreasing energy consumption than conventional protocols.
Direct transmissions and Minimum-transmission energy (MTE) are the conven-
tional protocols. In direct communication protocol, nodes are directly connected
to BS means they send data without any other sensor node. The nodes which are
far away from BS die earlier because they require large amount of transmission
energy due to large distance from BS. This protocol is acceptable only if BS is
close to the nodes. In MTE, nodes send data to BS through intermediate nodes.
These nodes provide a routing path for other nodes and also sense the environ-
ment. The intermediate nodes are selected by using transmit amplifier energy. If
there are three nodes A,B and C in network, the transmission is done by node A
to C through B if and only if this condition fulfils [1].
ETx(k, d = dAB) + ETx(k, d = dBC) < ETx(k, d = dAC) (3.1)
The transmission energy from A to B and from B to C is less than the transmission
energy of nodes A to C then the node B will be elected as intermediate node. In
MTE the nodes which are nearby to BS will die more quickly than the nodes which
are far away. Nearby nodes are used as intermediate nodes which consume more
energy than other nodes.
In LEACH, radio interference is produced when multiple clusters transmit at a
same time. To reduce this interference, each CH uses CDMA code and transmit
data to its members. They transmit data in an order which reduces the interfer-
ence.
3.1.2 TEEN
TEEN is a clustering routing protocol and related to reactive network. It is an
extension of LEACH protocol using same clustering scheme with an extra entity
which is a threshold value. TEEN protocol is data centric in which nodes get data
9
on certain parameters such as temperature, pressure, velocity and humidity e.g if
the node sense temperature greater than 50 then it sends data to BS otherwise
remain in sleep mode. If adjacent nodes have similar data then it is more favorable
to aggregate the data from each node rather than every node sends it separately.
The main features of TEEN protocol are that nodes have to transmit to their CH
to dissipate less energy, additional computation is done only by CH to save energy,
CHs present at high level of hierarchy have to transmit data which consume more
energy. To overcome this problem, all nodes will be CH for a time period T (cluster
Period).
In TEEN, nodes sense environment all the time and transmission is done only when
there is a drastic change. Due to minimal transmission less energy is consumed.
CHs broadcast two threshold values to its members which are,
• Hard threshold
• Soft threshold
Hard threshold is the absolute value, when a node sense this value transmitter is
switched on and send data to the CH. A small change in the sensed value is soft
threshold. Node sense the environment, when hard threshold is achieved node
transmits the sensed data. Hard threshold decreases the transmission of data
because nodes only transmit when they sense the drastic change. Soft threshold
also reduces the number of transmission by ignoring small changes in sensed value.
TEEN protocol is applicable in time critical data sensing applications, where user
needs the data instantaneously. Data transmission consumes more energy so,
nodes only sense the environment and transmit only when they sense a threshold
value. The Drawback of TEEN protocol is that nodes will never communicate if
the threshold value is not reached, the notification will not received by user even
if all nodes die. It is implemented where no collision exists. TDMA scheduling
can be used to avoid this problem but it introduces the delay. This scheme is not
suitable where user needs the data in periodic way.
CH selection criteria is same in both TEEN and LEACH protocol. The decision
is made by a random number selection between 0 and 1, member nodes will be
CH if selected number is less than the given threshold euation [1],
T (n) =
p
1−p(rmod 1p)
if nϵG
0 otherwise(3.2)
10
Where P is the percentage of CHs, r is the current round and G is the set of
nodes which are eligible to b CHs. 1pare the EPOCH rounds after which a node
is eligible to b a CH. When it is selected, data is transmitted to BS in compressed
form.
3.1.3 DEEC
DEEC is designed for heterogeneous network in which some nodes are advanced
with more energy than normal nodes. All the nodes are equipped with different
energy levels to increase the network lifetime. All nodes have same energy in
initial level, by reenergizing the sensor nodes either by adding nodes which have
high energy or to provide energy to already exists nodes. The energy nodes having
low energy will die earlier than other nodes. CH selection is based on the remaining
energy of node and average energy of the network. The data is gathered by CHs
from its node members and forward to the BS. EPOCH is different for each node
according to its residual and initial energy. The nodes having high residual and
high initial energy will be CHS more time than with low energy nodes. DEEC is
used to increase the network life time.
In DEEC, there are two type of nodes such as advanced and normal nodes but
DEEC is multi level heterogeneous network. The total initial energy of this net-
work is given by,
Etotal =N∑i=1
E0(1 + ai) = E0(N +N∑i=1
ai) (3.3)
In DEEC, there are some nodes with extra energy represented by a and m is the
probability of extra nodes. CHs election is based on the ratio between residual
energy of each node and average energy of the network. epochs of being cluster
head is different according to their initial and residual energy. The nodes having
high initial and residual energy will have more chances to become cluster head [4].
CH selection is based on the threshold equation [4],
pi = popt[1−E(r)− Ei(r)
E(r)] = popt
Ei(r)
E(r)(3.4)
Where popt is the reference value of the average probability of Pi, ¯E(r) is the
11
average energy of the network and Ei(r) is the residual energy of the network at
round r. Total number of CHs per round per epoch is,
N∑i=1
pi =N∑i=1
poptEi(r)
E(r)= popt
N∑i=1
Ei(r)
E(r)= Npopt (3.5)
T (si) =
pi
1−pi(rmod 1Pi
)if siϵG
0 otherwise(3.6)
Threshold value is given in above equation on which CH election criteria is based.
In two level heterogenous network the value of popt is given by,
padv =popt
1 + am, pnrm =
popt(1 + a)
(1 + am)(3.7)
Then use the above padv and pnrm instead of popt in above equation for two level
heterogeneous network as supposed in equation (3.8),
pi =
poptEi(r)
(1+am)E(r)if si is the normal node
popt(1+a)Ei(r)
(1+am)E(r)if si is the advanced node
(3.8)
It can be extended to multi level heterogenous network which is given as;
pmulti =poptN(1 + ai)
(N +∑N
i=1 ai)(3.9)
Above pmulti is used instead of popt to get pi for heterogeneous node. pi for the
multilevel heterogeneous network is given as;
pi =poptN(1 + a)Ei(r)
(N +∑N
i=1 ai)E(r)(3.10)
12
In DEEC we estimate average energy E(r) of the network for any round r as in;
E(r) =1
NEtotal(1−
r
R) (3.11)
R denotes total number of rounds and is estimated as follows;
R =Etotal
Eround
(3.12)
If the residual energy of the nodes is higher than average energy of network than
it have more chances to become CH.
3.1.4 H-TEEN
H-TEEN is a variant of TEEN protocol, introducing a hierarchy of clustering to
better cope with large network area. When number of layers in hierarchy is small
TEEN consumes lot of energy because of larger distance so, H-TEEN performs
better due to less consumption of energy in large network. H-TEEN is a 4 layer
hierarchal clustering where sensors self organize into clusters and build a tree of
transmissions and propagate data to the CH. CH selection is same as in TEEN
and LEACH. The threshold equation is given as [5],
T (n) =
pc
1−pc(rmod 1pc
)if nϵG
0 otherwise(3.13)
Here r is the current round and G is the set of nodes eligible to become CH.
The threshold value increases as rounds pass, alive nodes become a CH after
rounds 1p. When a node become CH it broadcasts an advertisement message to
all member nodes. Nodes receive the message and elect the CH which depends
on signal strength. When node decides cluster, it transmits a message to CH
belonging to that cluster. This process is done by using CSMA MAC protocol
to avoid collisions. The CH broadcasts a TDMA schedule for transmissions to
all its members. Now the next level of hierarchy is build. Previous CHs decides
whether they could be CH for next level of hierarchy. If they are eligible then
sends an advertisement message otherwise another node is selected to be CH.
13
After clustering, data transmission starts from nodes to CH. TDMA schedule is
broadcasted to all member nodes for transmission. To reduce interference CDMA
code is used, each node choose a different code. Moreover, two threshold values
are used, the hard and soft threshold values. Nodes sense the field continuously,
when hard threshold is reached node transmit data to BS. Soft threshold is small
change in that sensed value.
3.1.5 CAMP-TEEN
CAMP-TEEN is the extension of TEEN protocol, most suitable for the applica-
tion of landslide prediction. Nodes sense the slight movement of soil and change
in parameters that occur before land slide. CAMP enhance localization and en-
ergy efficiency of multi-hop routing protocol and TEEN is an extended version
of LEACH which saves energy by using threshold values. It is useful in landslide
prediction applications because each rock have different threshold values.
In CAMP-TEEN, one node broadcasts a beacon pulse. Nodes which are nearby
to that node, receive this beacon and sends an acknowledgement return to beacon
node. The acknowledgment has the distance between nodes and beacon node based
on RSSI (Received signal strength indication). It constructs the neighborhood
table for each node until all nodes have their neighboring table. CAMP uses
distributed clustering in which CH is selected on the basis of local information of
nodes. In CAMP-TEEN, CH selection criteria depends on a timer which is given
as [6] ,
T (v) =K
E− α (3.14)
Where K is the proportionality constant which is taken as 1, E is the normalized
energy of the node and α is the random number between 0 and 1. Timer starts for
every nodes by using above equation. The node with least timer value will have
high energy as they are inversely proportional to each other. The high energy
node will be elected as a CH then the nodes in neighboring of CH will terminate
their timers. CHs broadcast TDMA schedule to their cluster members. Nodes
transmit data to CH, it collects the data and forwards it to BS.
In table (1), some features of protocols are described in terms of their types of
communication, network and routing. Moreover, their CH selection criteria is also
mentioned.
14
Protocols
NetworkType
Com
municationType
Rou
tingType
CH
Selection
Criteria
DEEC
Heterogeneous
Single-hop
Proactive
Average
energy
ofnet-
workan
dresidual
en-
ergy
ofnode
LEACH
Hom
ogeneous
Single-hop
Proactive
Thresholdbased
prob-
ability
TEEN
Hom
ogeneous
Multi-hop
Reactive
Thresholdbased
prob-
ability
H-T
EEN
Hom
ogeneous
Multi-hop
Reactive
Thresholdbased
prob-
ability
CAMP-T
EEN
Hom
ogeneous
Single-hop
Reactive
Tim
erbased
probab
il-
ity
Table
3.1:FeaturesofDEEC,LEACH,TEEN
anditsVariants
15
Chapter 4
Sink Mobility
In WSN, sensors are deployed to transmit data and sink is placed in the network
anywhere to aggregate that data. Sink may be static or mobile, it depends on
application requirements. Research has shown that energy of nodes near to sink
exhausted very quickly in WSN where the sink is fixed, as a result networks get
disconnected. To overcome this problem and prolong the life time of network
mobile sink is used to collect the data. Data is transmitted in two ways either in
push mode or in pull mode. In push mode sensors send data without any request
from sink and in pull mode, sensors transmit only on sink’s request. Sensors which
are nearby to sink consume more energy because they share more sensor to sink
paths in case of static sink. This leads uneven energy depletion and produces
energy holes which shortens the network lifetime.
4.1 Data Aggregation Process in Sink Mobility
Sink mobility is implanted in many applications such as patient monitoring, land
sliding prediction and water monitoring. There are two process by which data can
be aggregated named as Direct-Contact Data Collection and Rendezvous based
data collection. In first process, mobile sinks aggregate data directly from sensors
by visiting all sensors to maximize energy savings. It reduces the message relay
overhead sensors and enhance energy savings. When sink is moving in slow speed,
it can aggregate maximum data. In Rendezvous based data collection, sink visits
only few points to reduce delay. It collects data from some sensors which avoid
the large distance at multi hop communication[27].
16
4.2 Strategies of Sink Mobility
Sink changes its position randomly according to the requirement of mobility. There
are many ways in which sink can move as if there is a network of 100m × 100m,
it can move on the top of network, bottom of network, diagonally in network
and in many other directions. There are three sink mobility strategies in which
sink can move randomly in the network. In residual based energy strategy, sink
moves toward the region of center residual energy of cluster which balances energy
consumption. The center residual energy is the position of cluster nodes according
to residual energy. Each cluster member sends its residual energy to sink which
evaluates the energy of node according to requirement. In event based strategy,
event region will be main area where sink will always move. Event region is where
cluster have maximum data flow. It reduces the transmission path and increases
network lifetime. Another strategy is combination of both residual based and
event based strategy to obtain a Hybrid strategy. The sink first moves toward
an event center then towards energy center. It will be connected to one of them
always which results in larger energy gain[27].
Sink changes its position randomly in network. Before the sink changes its posi-
tion, it stops for a fixed amount of time to aggregate the data from sensors within
its range called pause time. During pause time, the sink broadcasts a beacon frame
to its neighboring node for transmitting the data packets. When node sends the
data, sink broadcasts another beacon frame to stop transmission which reduces
the packet drop. The network life time can be extended if mobile sink balances the
traffic load of nodes. To minimize the traffic load shortest path routing is used.
For increasing the network life time in applications where the delay tolerant is an
important factor, a network model is proposed known as Delay Tolerant Mobile
Sink Model (DT-MSM) [9]. In this model, nodes can reschedule the transmission
for appropriate time. When the node is not in the range of mobile sink it stores
the data. In this way, it reduces the packet drop and energy of the node.
4.3 Model for Sink Mobility
WSNs are based on large number of tiny sensor nodes which have limited energy.
WSNs applications are in medical, environment monitoring and also in security
surveillance. Strive for the betterment of WSNs is still motivating to work on
different techniques and modification, to enhance the life time and low energy
consumption. The main constraint in the study of WSNs is the lifetime due to
17
limited energy of tiny sensors. Lot of work has done and yet many doors are still
open to explore. Purpose of study is to maximize the lifetime of WSNs without
affecting the cost of network. Mainly focus of the paper is on sink mobility and
its motion on different trajectories. In [28] speed of the moving sink is discussed,
its fast or slow in both cases lifetime of the network will improve because mobility
increases the dimension(degree of freedom) of the problem. Existing routing pro-
tocols related to homogeneous and heterogeneous networks are used and mobile
sink is introduced in the field. Study is based on data collection in WSNs by
considering sink mobility and as well as routing protocol.
4.3.1 Preliminaries
In [28], authors selected system model with digraph G(V,E), where V are vertices,
E are edges and c : V × V −→ ℜ+. Properties of the system are same with few
assumptions. i) All the sensor nodes are stationary, only sink is changing its
position. ii) Sink location is within a finite range Λ. iii) Sink is able to transmit
the sensed data to long rang destinations from the considered WSN. iv) Energy
consumption in the WSN is only during reception and transmission.
4.3.2 System Model
AWSN model is proposed in which V is the number of sensor nodes which will stay
awake in whole network lifetime and will sense and aggregate data from the field.
xij is the cost of flow going from vertex i to j. Network lifetime is defined as last
node will die. This model is designed for both homogeneous and heterogeneous
networks. The only energy consumption is in transmission and reception of the
data.
E = Etij + Er
ki (4.1)
where Etij and Er
ki are the energies of transmission and reception, respectively.
Strive is to improve the lifetime of the network by introducing two types of nodes
normal and advance, and making the sink mobile. In this model sink is mechani-
cally driven and can be recharged, so energy is not a constraint on moving sink. I
have exploited the sink mobility on all protocols, and compare their results here.
Sink is moving on the top of network. Moving sink will be collecting data, not
randomly but on the defined path. To avoid buffering over flow of the information
18
packets received at nodes, the tour of the sink and its sojourn locations have spe-
cific time, so that all nodes in the network can easily transfer their data without
loss. To make it cost-effective sojourn tour is predefined and the distance between
the two locations is bounded by rmax. Observed that sum of the sojourn locations
is actually the network lifetime.
4.3.3 Persuade Problem
In this section, I will discuss General assumptions about WSN. Later some math-
ematical equations supporting our proposed model. For convenience we assume
that the set of n sensor nodes are deployed randomly in the network. Similar to
[25] xij is the transmitted data rate from node i to j.
Etij = Ct
ij.xij (4.2)
here Ctij is the required energy to transmit one unit of data from node i to node
j. The energy consumed at node i per unit of time for receiving data from data
from node k is given by,
Erki = γ.xij, (4.3)
γ is a given constant. At node i, total energy consumption per unit time is:
∑i∈N
Etij +
∑k∈N
Erki =
∑i∈N
Ctij.xij +
∑k∈N
γ.xij (4.4)
In this model advance, normal and other nodes with high energy are used. During
the cycles of clustering we observed that the normal nodes will die first as compared
to other high nodes. Because the energy of advance nodes is greater than normal
nodes and have greater probability of becoming CH. Otherwise there will be no
effect on the processing of data in the network.
This model can be expressed by linear programming as:
Objective: Maximization of life time
Maximize T =k∑
i=1
ti (4.5)
19
subject to:
T.(∑i∈N
Ctij.xij +
∑k∈N
γ.xij) ≤ E ∀i, j, k (4.6)
n∑i=1
xij =n∑
i=1
xji ∀i, j (4.7)
xij ≤ Rij.ti ∀i, j (4.8)n∑
i=1
x0,i =n∑
i=1
xi,0 = 1 ∀n (4.9)
n∑i=0
n∑j=0
dijxij ≤ L (4.10)
dijxij ≤ rmax ∀i, j (4.11)
xij ≥ 0 ∀i, j (4.12)
Rij ≥ 0 ∀i, j (4.13)
ti ≥ 0 ∀i (4.14)
Equation (4.6) is representing energy constraint. The sum of the energy node
i is consuming in transmission and reception of the data in the whole network
lifetime is bounded by initial energy E. Equation (4.7) is flow conservation in all
locations of sink. Equation (4.8) is the rate constraint which is explaining that
total information rate which is flowing through the link (i, j) should not exceed the
link capacity Rij which is the upper bound of the transmission rate (bytes/second).
Equation (4.9) and (4.10) states that starting and ending location of sink is same.
Equation (4.11) implies that the distance completed by the sink is not greater
than L. Equation (4.12) represents the distance between two consecutive locations
of the sink is not more than rmax.
This problem can be mapped in other way round in terms of energy.
Object function: Minimization of energy consumption
Minimize E =∑k:i∈n
pkij.ti (4.15)
subject to:
qixij ≥ 1 ∀i, j (4.16)∑k:i∈n
pkij.ti ≤ E ∀i, j, k (4.17)∑i,j
∑k
pk,lij ≤ Pi, ∀i, k (4.18)
fk,lij ≤ h(pk,lij ) ∀i, j, k, l (4.19)
20
pki is the power consumption of node i during kth epoch.
Equation (4.16), shows that the rate of information generation multiply with the
data flow coming on the node, during any epoch is at least 1. Equation (4.17),
sum of the energy consumed during every epoch is less than equal to initial energy,
energy conservation. pk,lij denotes the transmission power during kth epoch on sink
location l. Equation (4.18) is peak transmission power constraint Pi to node i
for all sink locations and time intervals. Equation (4.19) ensures capacity related
upper bound h(pk,lij ), where h is non decreasing concave function, on the achievable
link rate fk,lij on link (i, j) using power pk,lij .
4.3.4 Maximization of Network Lifetime
In this section lifetime of network is maximized by introducing sink mobility. I
presented the mathematical formulation for enhancing lifetime under proposed
framework. Two scenarios are compared by using static sink and mobile sink in
terms of network lifetime problem. Sensors are deployed randomly in the network.
Consider a WSN model G(V,E) in which N sensor nodes are deployed in a square
area with length D. V is the set of sensors and E is the set of links. When sensor
and sink is in transmission range, there will be a link. The static sink is deployed
in the center of network and sensor is mobile on the top of network. The data is
generated at the rate of di. The initial energy of nodes is Ei. The energy required
to transmit and receive data from node i to node j as above equations.
4.3.4.1 Network Lifetime of Static Sink Model
In this model, sink is deployed in the middle of the network. Sink aggregates the
data from nodes in both types of communication such as multi-hop and single hop
communication. di is the self originated data at node i.
Maximize T (4.20)
21
Figure 4.1: Sink Model
Subject to: ∑j∈N(ı)
xıȷ −∑
k∈N(ı)
xkı = di ∀i ∈ N (4.21)( ∑j∈N(ı)
Ctij.xij +
∑kı∈N(ı)
γ.xki
)T ≤ Ei ∀i ∈ N (4.22)
xij ≥ 0 (4.23)
T ≥ 0 (4.24)
Equation (4.21) is the flow conservation constraint which states that all outgoing
flows are equal to incoming flows. Constraint (4.22) is the energy constraint which
shows that total energy should be less than initial energy of node Ei.
4.3.4.2 Network Lifetime of Mobile Sink Model
In this model, sink changes its position on the top border of the network and
collects data from sensor nodes. Traveling time of sink is ignored here and l is the
possible locations on which sink will stop for a certain amount of time which is
named as pause time.
22
MaximizeN∑i=1
T (4.25)
Subject to:
T = T1 + T2 + T3 + .........Tl (4.26)∑j∈N(ı)
x(l)ıȷ −
∑kı∈N(i)
x(l)kı = di (4.27)
l∑l=1
( ∑j∈N(ı)
ctijx(l)ij +
∑k∈N(ı)
γ.x(l)ki
)≤ Ei (4.28)
N∑i=1
x1,l = xl,1 = 1 (4.29)
dijxij ≤ L (4.30)
Equation (4.27) and (4.28) are the flow and energy constraints. Constraint (4.29)
states that starting and ending location of sink is same. The length of one tour is
bounded by D which shows in equation (4.30). xij will always greater than zero
and incoming energy is greater than outgoing energy because there is some loss in
way.
4.3.4.3 Maximize Lifetime in Threshold based Protocol
TEEN is a threshold based protocol in which nodes sense environment continu-
ously and transmit only when there is a drastic change. CHs are elected by nodes
which broadcast two threshold values to its members such as hard threshold and
soft threshold. Hard threshold is the absolute value which is sensed by nodes and
start transmission. Soft threshold is the small change in the sensed value. These
values reduce the number of transmission and consume less energy. This concept
is implemented to maximize the network lifetime.
MaximizeN∑i=1
T (4.31)
23
Subject to:
N∑i=1
T = P li (4.32)
N∑i=1
P li = H.di (4.33)
N∑i=1
P li = ∆H.di (4.34)
N∑i=1
P li = S.di (4.35)
Where P li is the decision variable by which nodes decide whether to transmit or not.
H is the hard threshold value and S is the soft threshold value. Equation (4.33)
shows that transmission will be done only when nodes sense the hard threshold.
Equation (4.35) states that if node senses the small change in previous value, it
will transmit this change to CH.
4.3.4.4 Maximize Lifetime for scalable Model
The scalability is introduced to increase the network lifetime by deploying more
nodes. Number of nodes are increased by 100 to 200 but the area of network
remains same. The distance between nodes become shorter so the chances of
nodes to die become less which increases network lifetime.
Maximize
N∑i=1
T (4.36)
Subject to:
24
N∑i=1
si,j = d(i, j) (4.37)
2N∑i=1
si,j << d(i, j) (4.38)∑j∈N(ı)
x(l)ıȷ −
∑kı∈N(i)
x(l)kı = di (4.39)
|L|∑l=1
( ∑j∈N(ı)
ctijx(l)ij +
∑k∈N(ı)
γ.x(l)ki
)≤ Ei (4.40)
Equation (4.37) states that when nodes are 100, distance between two nodes is
equal to d(i, j). Equation (4.38) shows that when number nodes are increased,
distance between nodes will decrease. It will increase network life time because of
less energy consumption.
25
Chapter 5
Simulations and Results
To evaluate the performance of different homogeneous and heterogeneous proto-
cols, we have implemented it in MATLAB. The simulation has been executed on a
network of 100 nodes and a BS (fixed and mobile). In the network, nodes are de-
ployed randomly. The field size is 100m×100m and number of rounds are 5000 in
all scenarios. Our goal is to compare the performance of LEACH, DEEC, TEEN
and two extensions of TEEN protocol which are H-TEEN and CAMP-TEEN.
Here, three different metrics are used to analyze and compare the performance of
the protocols which are, total number of alive nodes, total number of dead nodes
and throughput in two different scenarios as,
• Sink is static.
• Sink is mobile (on the top of network).
5.1 Radio Dissipation Model
Figure 1 depicts the radio dissipation model, the energy expended by transmitting
L bit message over distance d is given as, [4]
ETx(L, d) =
L.Eelec + L.ϵfs.d2 if d ≤ d0
L.Eelec + L.ϵmp.d4 if d > d0
(5.1)
Where Eelec is energy dissipated to run the transmitter or receiver circuit, Efs
is free space transmit amplifier if dmax to BS < d0, Emp is multi path transmit
amplifier if dmax to BS > d0 and d is the distance between cluster members and
26
BS. Total energy dessipated during a round is given as [4],
Eround = L(2NEelec +NEDA + kϵmpd4toBS +Nϵfsd
2toCH) (5.2)
Where dtoBS is the distance between the CH and BS and dtoCH is the distance
between cluster members and CH [4].
dtoCH =M√2Πk
, dtoBS = 0.765M
2(5.3)
k =
√N√2Π
√ϵfsϵmp
M
d2toBS
(5.4)
Where K is the optimal number of clusters.
L bit packet Transmit
ElectronicsTx Amlifier
Receiver
Electronics
L bit packet
ETx(d)
EeleTX *L Eamp *L*d2
EeleRX *L
d
Figure 5.1: Radio Energy Dissipation Model
I have taken same energy model parameter for all the protocols which are given
in table 1.
27
Parameters ValuesNetwork field 100 m,100 mNumber of nodes 100E0 (initial energy) 0.5JData bit 4000Eelec 50 nJ/bitEfs 10 nJ/bit/ m2
Emp 0.013/pJ/bit/ m4
EDA 5 nj/bitPopt 0.1
Table 5.1: Values of Parameters
Figure 2 and 3 shows the homogeneous and heterogeneous networks in which
sensors are deployed randomly and BS is moving on the top of network.
Figure 5.2: Heterogeneous Network
5.2 Results
Figure 4 demonstrate network life time of all routing protocols in first scenario
when the BS is static and placed in center of the network. I observed that by
comparing LEACH, TEEN and DEEC the stability period of LEACH is shorter
almost 50% and 55% less than DEEC and TEEN because in LEACH, energy of all
nodes are same it takes no advantage of nodes that have more energy than other
nodes. The network life time of TEEN is better than LEACH and DEEC because
28
Figure 5.3: Homogeneous Network
in TEEN hard threshold and soft threshold values are used for nodes to transmit
the data. Nodes are in sleep mode in TEEN, they only transmit when sensed
the threshold value so, data transmission is done less frequently and less energy
is consumed. DEEC generates un-even number of CHs for every round that can
disturb performance of network where optimal number of CHs are necessary to
enhance networks life which are implemented in TEEN. When I consider H-TEEN
the nodes remain alive for maximum number of rounds more than 5000 because I
simulate it as a four level hierarchy. CHs at increasing level have to transmit data
to BS. When layers in hierarchy increases the transmission become shorter and
less energy is consumed. TEEN is two level hierarchy protocol and H-TEEN is
four level hierarchy protocol so, it performs better than TEEN and other protocols
for network lifetime.
Figure 5 demonstrate number of dead nodes as network operation proceeds. I
examine that first node of TEEN, LEACH, CAMP-TEEN, H-TEEN and DEEC
dies at 1386, 908, 200, 641, 1117. When we examine the dead nodes in figure 5,
it is obvious that in TEEN all nodes will die later than LEACH and DEEC. All
nodes die in LEACH,TEEN and DEEC at 3458, 5000, and 3000.
Figure 6 depicts the throughput from CHs to sink. The data rate of DEEC is
significantly greater than that of TEEN, LEACH, H-TEEN and CAMP-TEEN.
It means that cluster heads elected in case of DEEC are improved than other
protocols. In DEEC, nodes can transmit data continuously to BS while in TEEN
29
0 1000 2000 3000 4000 50000
10
20
30
40
50
60
70
80
90
100
Number of rounds
Num
ber
of a
live
node
s
LEACHTEENHTEENCAMPTEENDEEC
Figure 5.4: Number of alive nodes with static sink (100 nodes and 100m × 100m )
0 1000 2000 3000 4000 50000
10
20
30
40
50
60
70
80
90
100
Number of rounds
Num
ber
of d
ead
node
s
LEACHTEENHTEENCAMPTEENDEEC
Figure 5.5: Number of dead nodes with static sink (100 nodes and 100m × 100m )
30
and its variants, there is limited transmission because they are threshold based
protocol and have limited information to share with sink. Another reason is that
DEEC is a multi level heterogeneous protocol having some advanced and super
nodes which have extra energy than other normal nodes so they will die later than
normal nodes and they could be CHs more times than normal nodes.
0 1000 2000 3000 4000 50000
0.5
1
1.5
2
2.5
3x 10
4
Number of rounds
Pac
kets
tran
smitt
ed to
BS
LEACHTEENHTEENCAMPTEENDEEC
Figure 5.6: Throughput of protocols with static sink (100 nodes and 100m × 100m )
I compare all the protocols in second scenario where the BS is mobile on the top of
the network. When the distance between CH and BS is less, every CH sends data.
Figure 7 and 8 shows the network life time of all routing protocols. I examine that
due to mobility of sink, stability period of protocols decreases because when the
node is in the center of network it has equal distance to all nodes, they consume
same amount of energy. Now sink is on top of network, the nodes which are
at greater distance from sink will die quickly. Nodes consume greater energy at
greater distance so die earlier. The network life time increases in all protocols
because it enables to achieve less traffic load to nodes and reduces the delay in
transmission. When sink moves, CHs near to BS transmits the data so consumed
less energy and maximize the life time. It is more efficient in hierarchal clustering
based protocol.
Figure 9 shows the throughput of all proposed protocols. Less energy is consumed
31
0 1000 2000 3000 4000 50000
10
20
30
40
50
60
70
80
90
100
Number of rounds
Num
ber
of a
live
node
s
Mob−LEACHMob−TEENMob−HTEENMob−CAMPTEENMob−DEEC
Figure 5.7: Number of alive nodes with mobile sink (100 nodes and 100m × 100m )
0 1000 2000 3000 4000 50000
10
20
30
40
50
60
70
80
90
100
Number of rounds
Num
ber
of d
ead
node
s
Mob−LEACHMob−TEENMob−HTEENMob−CAMPTEENMob−DEEC
Figure 5.8: Number of dead nodes with mobile sink (100 nodes and 100m × 100m )
32
in this scenario because clustered node only transmit when there is minimum
distance between CH and BS. When the CHs are in range of sink they transmit
data directly. Due to minimum distance less energy is consumed than in fixed
base station. Throughput of Mob-DEEC is again better than all other protocols
in this scenario as in case of fixed BS.
0 1000 2000 3000 4000 50000
1
2
3
4
5
6x 10
4
Number of rounds
Pac
kets
tran
smitt
ed to
BS
Mob−LEACHMob−TEENMob−HTEENMob−CAMPTEENMob−DEEC
Figure 5.9: Throughput of protocols with mobile sink (100 nodes and 100m × 100m )
5.2.1 Analyzing Scalability
It is the main challenge to increase the number of nodes in existing network size.
By introducing scalability, routing protocols need to control some problems such
as link break in wireless link and other repair actions. When the number of nodes
and size of network is small these issues will not occur.
I introduce the scalability in all protocols to compare the performances. The per-
formance metrics are number of alive nodes, number of dead nodes and through-
put. Compare the performance of protocols with 200 nodes 100m× 100m dimen-
sion to evaluate the performance. When we increase the number of nodes and
dimensions are taken same then the performance of protocols are better because
sensors are densely deployed in same area of network. The distance between all
33
nodes become shorter, the chances of nodes to die become less which increases the
network lifetime.
0 1000 2000 3000 4000 50000
20
40
60
80
100
120
140
160
180
200
Number of rounds
Num
ber
of a
live
node
s
LEACHTEENHTEENCAMPTEENDEEC
Figure 5.10: Number of alive nodes with static sink (200 nodes and 100m × 100m )
Figures 10 and 11 shows the bar plot of throughput and alive nodes of five pro-
tocols. The average values of all protocols are found in terms of throughput and
number of alive nodes. The average throughput of LEACH, TEEN, HTEEN,
CAMPTEEN and DEEC in case of static sink are 11574, 15745, 14889 ,853,
22552. In case of mobile sink the average throughput of these protocols are 10771,
14582,16546, 948, 40872. The throughput of DEEC in both cases is better than
other protocols because of heterogeneous nature. By comparing network lifetime
of all protocols in both cases such as mobile sink and static sink it is concluded
that number of nodes alive in case of mobile sink is greater than static sink. The
average number of alive nodes of LEACH, TEEN, HTEEN, CAMPTEEN, and
DEEC are 27, 41, 59, 69, 41 in case of static sink. In case of mobile sink, the
average number of alive nodes are 25, 37, 59, 68, 39. The average number of alive
nodes are greater in TEEN and its variants because they are hierarchal clustered
protocols and have threshold values which decreases the energy consumption.
Firstly I compare all the protocols which elaborates that LEACH, TEEN and its
variants support homogeneous environment and DEEC is designed for heteroge-
34
0 1000 2000 3000 4000 50000
20
40
60
80
100
120
140
160
180
200
Number of rounds
Num
ber
of d
ead
node
s
LEACHTEENHTEENCAMPTEENDEEC
Figure 5.11: Number of dead nodes with static sink (200 nodes and 100m × 100m )
0 1000 2000 3000 4000 50000
0.5
1
1.5
2
2.5
3
3.5
4
4.5x 10
4
Number of rounds
Pac
kets
tran
smitt
ed to
BS
LEACHTEENHTEENCAMPTEENDEEC
Figure 5.12: Throughput of protocols with static sink (200 nodes and 100m × 100m )
35
0 1000 2000 3000 4000 50000
20
40
60
80
100
120
140
160
180
200
Number of rounds
Num
ber
of a
live
node
s
Mob−LEACHMob−TEENMob−HTEENMob−CAMPTEENMob−DEEC
Figure 5.13: Number of alive nodes with mobile sink (200 nodes and 100m × 100m )
0 1000 2000 3000 4000 50000
20
40
60
80
100
120
140
160
180
200
Number of rounds
Num
ber
of d
ead
node
s
Mob−LEACHMob−TEENMob−HTEENMob−CAMPTEENMob−DEEC
Figure 5.14: Number of dead nodes with mobile sink (200 nodes and 100m × 100m )
36
0 1000 2000 3000 4000 50000
1
2
3
4
5
6
7x 10
4
Number of rounds
Pac
kets
tran
smitt
ed to
BS
Mob−LEACHMob−TEENMob−HTEENMob−CAMPTEENMob−DEEC
Figure 5.15: Throughput of protocols with mobile sink (200 nodes and 100m × 100m)
0 2 4 6 8 10 120
0.5
1
1.5
2
2.5
3
3.5
4
4.5x 10
4
Pac
kets
tran
smitt
ed to
BS
LEACHMob−LEACHTEENMob−TEENHTEENMob−HTEENCAMPTEENMob−CAMPTEENDEECMob−DEEC
Figure 5.16: Throughput of all protocols showing in bar graph
37
0 2 4 6 8 10 120
10
20
30
40
50
60
70
Aliv
e no
des
LEACHMob−LEACHTEENMob−TEENHTEENMob−HTEENCAMPTEENMob−CAMPTEENDEECMob−DEEC
Figure 5.17: Number of alive nodes showing in bar graph
Protocols Static Sink Mobile SinkThroughput Alive Nodes Throughput Alive Nodes
DEEC 1 4 1 4TEEN 3 2 3 2LEACH 4 3 4 3H-TEEN 2 1 2 1CAMP-TEEN 5 5 5 5
Table 5.2: Comparing Performance of DEEC, LEACH, TEEN and its Variants
38
neous network. DEEC is more scalable than other protocols and consume less
energy. H-TEEN is highly energy efficient due to its hierarchy. Then we compare
the two scenarios, fixed BS and mobile BS. I examine that results with mobile BS
is better due to increasing network life time and throughput.
5.3 Conclusion
Energy efficiency and network lifetime are challenging issues in WSNs. Clustering
technique have been proposed to resolve these issues by using different clustering
schemes. First, I compare all routing protocols in terms of number of alive nodes,
number of dead nodes and throughput. I conclude from our analytical simulations
that DEEC performs better in sending packets among all protocols. H-TEEN is
more energy efficient because of hierarchial clustering and threshold value. Then,
I introduce the mobility of sink in all proposed protocols to enhance the network
life time and compare their performances. In last, I introduce the scalability in all
protocols and evaluate their results.
39
Bibliography
[1] Wendi Rabiner Heinzelman, Anantha Chandrakasan and Hari Balakrish-
nan ”Energy-Efficient Communication Protocol for Wireless Microsensor Net-
works”.
[2] Arati Manjeshwar and Dharma P. Agarwal ”TEEN: A routing Protocol for
Enhanced Efficiency in Wireless Sensor Networks” . 53
[3] Georgios Smaragdakis, Ibrahim Matta and AzerBestavros” SEP: A Stable
Election Protocol for clustered heterogeneous wireless sensor networks”.
[4] L. Qing, Q. Zhu and M. Wang, ”Design of a distributed energy-
efficientclustering algorithm for heterogeneous wireless sensor networks”. EL-
SEVIER, Computer Communications 29, 2006.
[5] S. Nikoletseas , I. Chatzigiannakis, H. Euthimiou ,A. Kinalis , A. Antoniou
and G. Mylonas ”Energy Efficient Protocols for Sensing Multiple Events in
Smart Dust Networks”.
[6] Kalyana Tejaswi, Prakshep Mehta, Rajat Bansal, Chandresh Parekh, S. N.
Merchant and U. B. Desai ”Routing Protocols for Landslide Prediction using
Wireless Sensor Networks”.
[7] Vivek Katiyar, Gopal Chand Gautam, Anil Kumar ”Improvement in LEACH
Protocol for Large-scale Wireless Sensor Networks”, ICETECT 2011.
[8] Arezoo Yektaparast, Fatemeh-Hoda Nabavi, Adel Sarmast ”An Improvement
on LEACH Protocol (Cell-LEACH)”,Department of Computer Engineering,
Science and Research Branch, Islamic Azad University, Khouzestan, Iran,
ICACT 2012.
[9] Mohammad Javad Hajikhani, Bahman Abolhassani, ”An Energy Efficient Al-
gorithm for Cluster-Head Selection in WSNs”, 2010 5th International Sympo-
sium on Telecommunications.
40
[10] ShenLin, Wang Liqin, Zhu Zhengwei, ”A Clustering Algorithm Based Ge-
ographic Location Information for Wireless Sensor Networks”, 2010 Interna-
tional Conference on Electrical and Control Engineering.
[11] Bilal Abu Bakr, Leszek Lilien, ”Extending Wireless Sensor Network Lifetime
in the LEACH-SM Protocol by Spare Selection”, 2011 Fifth International Con-
ference on Innovative Mobile and Internet Services in Ubiquitous Computing
[12] Arati Manjeshwar and Dharma P. Agrawal, ”APTEEN: A Hybrid Protocol
for Effficient Routing and Comprehensive Information Retrieval in Wireless
Sensor Networks”, Center for Distributed and Mobile Computing, ECECS
Department, University of Cincinnati, 2002
[13] Brahim Elbhiri, Saadane Rachid, Sanaa El fkihi, Driss Aboutajdine, ”De-
veloped Distributed Energy-Efficient Clustering (DDEEC) for heterogeneous
wireless sensor networks” IEEE 2010.
[14] Parul Saini, Ajay.K.Sharma, ” E-DEEC-Enhanced Distributed Energy Ef-
ficient Clustering Scheme for heterogeneous WSN”, 2010 first International
Conference on Parallel, Distributed and Grid Computing (PDGC).
[15] Parul Saini, Ajay.K.Sharma, ”Energy Efficient Scheme for Clustering Pro-
tocol Prolonging the Lifetime of Heterogeneous Wireless Sensor Networks”,
International Journal of computer Applications september, 2010.
[16] Z. Maria Wang, Stefano Basagni, Emanuel Melachrinoudis and Chiara Petri-
oli”Exploiting Sink Mobility for Maximizing Sensor Networks Lifetime”, Pro-
ceedings of the 38th Hawaii International Conference on System Sciences -
2005
[17] Wendi B. Heinzelman, Amy L. Murphy, Hervaldo S.Carvalho, Mark A. Per-
illo,Middleware to Support Sensor Network Applications, IEEE Network, pp.
6-14, 2004.
[18] Ioannis Chatzigiannakis, Athanasios Kinalis, Sotiris Nikoletseas ”Sink Mobil-
ity Protocols for Data Collection in Wireless Sensor Networks”.
[19] N. Vlajic, D. Stevanovic, ”Performance Analysis of ZigBee-based Wireless
Sensor Networks with Path-Constrained Mobile Sink(s).” 2009 Third Interna-
tional Conference on Sensor Technologies and Applications.
[20] Aristotelis Giannakos, Gregory Karagiorgos, Ioannis Stavrakakis, ”A
Message-Optimal Sink Mobility Model for Wireless Sensor Networks”, 2009
Eighth International Conference on Networks.
41
[21] M. Aslam, N. Javaid, A. Rahim, U. Nazir, A. Bibi, Z. A. Khan ”Survey
of Extended LEACH-Based Clustering Routing Protocols for Wireless Sen-
sor 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.
[22] 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
[23] Deepak Puthal, Bibhudatta Sahoo, Suraj Sharma, ”Dynamic Model for Effi-
cient Data Collection in Wireless Sensor Networks with Mobile Sink”.
[24] J. Luo, J. Panchard, M. Piorkowski, M. Grossglauser J-P Hubaux, Mo-
biRoute: Routing towards a Mobile Sink for Improving Lifetime in Sensor
Networks, 2nd IEEE/ACM Intl Conf. on Distributed Computing in Sensor
Systems (DCOSS), pp. 480-497, 2006.
[25] YoungSang Yun and Ye Xia ”Maximizing the Lifetime of Wireless Sensor
Networks with Mobile Sink in Delay-Tolerant”.
[26] A. Chakrabarti, A. Sabharwal, B. Aazhang,Using Predictable Observer Mo-
bility for Power Efficient Design of Sensor Networks, in The second Inter-
national Workshop on Information Processing in Sensor Networks (IPSN),
pp.129-145, 2003.
[27] Xu Li, Amiya Nayak, and Ivan Stojmenovic ”Sink Mobility in Wireless Sensor
Networks”.
[28] Jun Luo and Jean-Pierre Hubaux ”Joint sink mobility and routing to maxi-
mize the lifetime of wireless sensor networks: the case of constrained mobility”
IEEE/ACM TRANSACTIONS ON NETWORKING, JUNE 2010.
42