Available via license: CC BY 3.0
Content may be subject to copyright.
Research Article
Towards Network Lifetime Maximization:
Sink Mobility Aware Multihop Scalable Hybrid Energy
Efficient Protocols for Terrestrial WSNs
Mariam Akbar,1Nadeem Javaid,1Zahoor Ali Khan,2,3 Umar Qasim,4
Turki Ali Alghamdi,5Saad Noor Mohammad,1Syed Hassan Ahmed,6
Majid Iqbal Khan,1and Safdar Hussain Bouk6
1COMSATS Institute of Information Technology, Islamabad 44000, Pakistan
2Internetworking Program, FE, Dalhousie University, Halifax, NS, Canada B3J 4R2
3CIS, Higher Colleges of Technology, Fujairah Campus 4114, UAE
4University of Alberta, Edmonton, AB, Canada T6G 2J8
5College of CIS, Umm Al-Qura University, Makkah 21955, Saudi Arabia
6Kyungpook National University, Daegu 702-701, Republic of Korea
Correspondence should be addressed to Nadeem Javaid; nadeemjavaidqau@gmail.com
Received April ; Accepted July
Academic Editor: Sana Ullah
Copyright © Mariam Akbar et al. is is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
We propose two routing protocols for Terrestrial Wireless Sensor Networks (TWSNs): Hybrid Energy Ecient Reactive (HEER)
and Multihop Hybrid Energy Ecient Reactive (MHEER) routing protocol. e main purpose of designing these protocols is
to improve the network lifetime and particularly the stability period of the underlying network. In MHEER, the node with the
maximum energy in a region becomes cluster head (CH) of that region for that particular round (or cycle) of time and the number
of the CHs in each round remains the same. Our techniques outperform the well-known existing routing protocols: LEACH, TEEN,
and DEEC in terms of stability period and network lifetime. We also calculate the condence interval of all our results which helps
us to visualize the possible deviation of our graphs from the mean value. We also implement sink mobility on HEER and MHEER.
We refer to them as HEER-SM and MHEER-SM. Simulation results show that HEER-SM and MHEER-SM yield better network
lifetime and stability region as compared to the counterpart techniques. We have also carried out simulations with and
nodes in the same eld dimensions besides nodes. Simulations prove that the proposed schemes show the same behavior with
and nodes; that is, HEER and MHEER are scalable as well.
1. Introduction
A Wireless Sensor Network (WSN) consists of a number of
tiny wireless sensors dispersed throughout the network area.
ese sensors are very small in size and their basic function is
to monitor any particular environment. A WSN can be used
for security purposes, medical applications, environmental
monitoring, and so forth. ese sensor nodes monitor their
environment and send the desired data to the base station
(BS) via some routing protocol. As long as a sensor does not
run out of power, it keeps sending its data to the BS. But a
sensor cannot be recharged from time to time. Whenever its
energy is completely consumed, it is no longer able to sense
and send its data. So it is very important to implement an
ecient routing protocol to improve the network lifetime and
particularly its stability period. e network lifetime is the
time period when a network starts working until the last node
dies. On the other hand stability period is dened as time
period from the start of a network till the death of very rst
node in the network. Very less energy is consumed in sensing
Hindawi Publishing Corporation
International Journal of Distributed Sensor Networks
Volume 2015, Article ID 908495, 16 pages
http://dx.doi.org/10.1155/2015/908495
International Journal of Distributed Sensor Networks
data or aggregating it as compared to the energy consumed in
transmission or reception of data. So a routing protocol plays
a vital role in improving lifetime of a WSN.
Many protocols use clustering as their routing scheme [–
] as this technique is very eective for data transmission
in WSNs. In this technique, the member nodes of a cluster
select a CH among themselves for a particular round. All
the cluster members send data to their respective CH. e
CH receives that data, aggregates it, and then sends it to the
BS. Aggregation gets rid of redundant data and only useful
data is sent to base station which saves energy. Clustering
can be done in two types of networks, that is, homogeneous
and heterogeneous. In homogeneous WSNs, all nodes have
thesameenergylevel,whereasnetworkswithdierent
node energy levels are termed as heterogeneous networks.
Multihoping between CHs is also a technique used for the
extension of lifetime of large scale networks [].
Protocols can be classied as proactive and reactive.
When the nodes periodically send their data to the BS, these
are referred to as proactive. ese protocols send information
of relevant parameters aer a xed period of time. ese
types of networks are usually used for applications requiring
periodic data monitoring. When the nodes react immediately
to sudden and drastic changes in the value of the interested
parameter then the protocols are said to be reactive. In
reactive protocols, node does not have to wait for a xed
period of time to sense and transmit the data. Its sensors
switch on their transmitters whenever there is a drastic
change in the value of interested parameter. ese protocols
are suited for time critical applications.
Our proposed protocol is a reactive one for homogeneous
networks which uses initial energy and residual energy of a
node for selecting a CH. Aer the selection of a CH, it will
broadcast two threshold values. e transmission occurs if
and only if the Current Value reaches the threshold value. is
techniquereducesthenumberoftransmissionsandprolongs
thenetworklifetimeandstabilityperiod.
Clustering may be static or dynamic. In static clustering,
the clusters do not change their size, whereas in dynamic
clustering, depending upon the network characteristics, the
clusters change their size during their lifetime period.
Our scheme is based on static clustering. Whole area is
divided into regions. Each of these regions acts as a
cluster. Nodes are randomly deployed in each region and only
asinglenodecanbecomeaCHineachregionforaparticular
round. BS is located at the centre of the whole network area.
Nodes send their data to the CH of their region via direct
communication. e data from the CH to the BS is com-
municated through direct communication or multihoping
depending upon its location. e CHs close to the BS send
their data to the BS by using direct communication, whereas
nodes which are farther from the BS send their data using
Multihop Transmission between them and the CHs which are
closer to the BS. e results suggest that it further enhances
thenetworklifetimeandstabilityperiod.
Since communication distance has a signicant impact
on the energy consumption cost of nodes, we also implement
sink mobility in our protocols to reduce the communication
distance. In other words, networks with mobile sink remain
alive for a longer period of time as compared to networks
without sink mobility. In sink mobility, the sink moves in
dierent locations of the network to collect the data. In our
protocols, sink does not collect the data during its motion.
It only collects data when it is at its sink locations in the
network. It stops at its sink locations and collects the data
from the nodes. ese sink locations are also referred to
as sojourn locations. e results depict that HEER-SM and
MHEER-SM yield better network lifetime and stability region
as compared to HEER and MHEER, respectively. It is worth
mentioning here that this work is extended form of the work
in [].
2. Related Work
Many researchers have reviewed and analyzed the perfor-
mance of dierent protocols in WSNs [,]. LEACH []
is the rst hierarchical clustering algorithm for WSNs. It is
based on the dynamic clustering technique. Aer certain time
period, nodes are organized into clusters and each CH is
selected on the basis of probability. Due to cluster formation,
the distance between CH and member nodes is reduced.
Nodes transmit their data at minimum communication
distance to minimize the energy consumption cost. is
increases the network lifetime as well as throughput of the
network. LEACH outperforms classical clustering algorithm
by using adaptive clustering and rotating CHs. is saves
energy as transmission will only be performed on that specic
CH rather than all the nodes.
reshold Sensitive Energy Ecient Network (TEEN) []
was proposed by Manjeshwar and Agrawal in . It is a
reactive protocol for time critical applications. Its CH selec-
tion and cluster formation of nodes are the same as those of
LEACH. In this scheme, CH broadcasts two threshold values,
thatis,Hardreshold(HT)andSoreshold(ST).HT
is the absolute value of an attribute to trigger a sensor node.
HT allows nodes to transmit the event, if the event occurs
in the range of interest. erefore, this not only reduces the
number of transmissions but also increases network lifetime.
TEEN is designed for time critical application; nodes only
transmit data when it is needed according to HT. In the
remaining time they switch o the transmitter and get active
when HT arrives. e disadvantage of this scheme is that
the network could not get operational until HT arrives. If
network does not observe HT, user will not receive ant data
from the network and even no information whether any node
is alive.
Smaragdakis et al. [] proposed a two-level heterogeneous
aware protocol, consisting of normal and advanced (high
energy) nodes. It is based on the weighted election prob-
abilities of each node according to their respective energy
to become a CH. Intuitively, advanced nodes have more
probability to become a CH than normal nodes, which seems
logical according to their energy consumption. Stable Elec-
tion Protocol (SEP) does not require any global knowledge
of the network. e drawback of SEP is that it does not
consider the changing residual energy of the node; hence, the
probability of advanced nodes to become CH remains high
irrespective of the residual energy le in the node. Moreover,
International Journal of Distributed Sensor Networks
SEP performs below par if the network is more than two
levels.
In , Qing et al. []proposedDistributedEnergy
Ecient Clustering (DEEC) protocol for WSNs. is scheme
minimizes the energy consumption of the nodes by consid-
ering average energy of the network and uses it as a reference
energy. Due to this approach global knowledge of energy
is not required. DEEC is a clustering protocol for two and
multilevel heterogeneous networks. In DEEC the probability
for a node to become CH is based on residual energy of the
node and average energy of network. e epoch for nodes to
become CH is set according to the residual energy of a node
and average energy of the network. e node with higher
initial and residual energy has more chances to become a CH
than the low residual energy node. DEEC performs well in
multilevel heterogeneous WSN as compared to LEACH and
SEP.
Ecient Scheduling for the Mobile Sink in Wireless
Sensor Networks with Delay Constraint (ESWC) is proposed
by Gu et al. in [].isprotocolimplementssinkmobility
to improve the network lifetime. It also bounds the delay
caused by the movement of the sink. A general and practical
unied formulation is also provided in this scheme that
analyzes jointly the sink mobility, routing, and delay of the
network. e authors also propose polynomial-time optimal
algorithm. ey compare the advantages of mobile sink in
the network with that without mobile sink. is protocol also
discusses dierent sink trajectories and their eects on the
lifetime, delay, and throughput.
Also in [], the authors implement the sink mobility
technique to improve the network lifetime and the stability
region. As the mobile sink is driven by petrol or electricity.
is protocol also bounds the travel distance of mobile sink to
avoid data loss during the transition of mobile sink between
sink locations. When mobile sink stops at a certain sink
stop, routing tree is constructed and it causes overhead. To
avoid it sink stops at a stop for a denite amount of time
on each stop. e authors in this paper dened that the
sojourn trip of a mobile sink is the sum of sojourn times
in the trip. e authors rst formulated the problem as a
Mixed Integer Linear Programming (MILP), with objective of
maximizing the sum of sojourn times in the whole trip. Due
to its NP-hardness, they then devised a novel heuristic for it.
en they conducted extensive experiments by simulations to
evaluate the performance of the proposed algorithm in terms
of network lifetime.
e authors in [] also improved the network lifetime
byjointlyconsideringsinkmobilityaswellasroutingby
considering sink to the nite locations. ey also proved
the NP-hardness of their proposed model implementing
multiple mobile sinks. ey proved the NP-hardness of the
problem and also investigated the induced subproblems. ey
developed an ecient primal-dual algorithm to solve the
subproblem involving a single sink; then they generalized
this algorithm to approximate the original problem involving
multiple sinks. Finally, they applied the algorithm to a set
of typical topological graphs; the results demonstrate the
benet of involving sink mobility, and they also suggested the
desirable moving traces of a sink.
In WSNs, sink mobility balances the nodes energy con-
sumption. Nodes have to reconstruct the routes for data
transmission when mobile sink moves towards next stop.
During transition time, data dissemination is a challenging
task. In [], authors proposed a Virtual Grid Based Dynamic
Routes Adjustment (VGDRA) scheme. It reduces the route
reconstruction cost of nodes. For this purpose they optimize
the sink location and also dene communication rules. Few
nodes reconstruct their routes to readjust the path with the
sink. rough this scheme they extend the network lifetime.
In [], authors proposed a Lifetime Optimization Algo-
rithm with Mobile Sink Nodes for WSNs based on location
information (LOA MSN). For obtaining the location infor-
mation of nodes authors used satellite and RSSI positioning
algorithms. ey established the movement paths with the
help of lifetime optimization and path selection models. Sink
obtains the location information of the nodes. en, through
graph theory model, they obtain the movement paths. e
mobile sink gathers data from the nodes in the grid center.
rough experiments, they show that sink nds optimal
path and minimizes nodes’ energy consumption cost, which
leads to prolonged network lifetime. LOA MSN uses multiple
mobile sinks and minimizes the energy consumption cost;
however, it increases data gathering latency.
In the paper [], authors presented an energy ecient
routing scheme that maximizes the network throughput. For
data forwarding they use multilayer clustering design that
nds forwarder node. e role of CH rotated among the
nodesisbasedonthethresholdvalues;thisreducesthe
number of packets dropped. ey use Cluster Designing
Algorithm (CDA) architecture for the selection of forwarder
node and inter- and intracluster routing, CH rotation, and
the data delivery; all these processes are energy-aware.
e experiments show that careful selection of forwarder
node leads towards energy ecient routing in intracluster
and interclusters. It also increases throughput and network
lifetime. It is also concluded that CH rotation in each round
consumes energy; rather CH works until it consumes a
certain amount of energy. Aer that other suitable nodes take
the charge of CH.
Authors in [] propose a scheme to improve throughput
of the network by considering base station placement prob-
lem for Wireless Sensor Networks with Successive Interfer-
ence Cancelation (SIC). rough mathematical model they
address this issue. is model is useful to identify a necessary
condition for SIC by considering distances from sensor nodes
to the base station. To achieve this they divide the network
eld into feasible regions and select a point in each small
region for the stop of base station. e small region with the
greater throughput is considered as a solution.
3. Counterpart Protocols in Brief
is section provides the readers with brief discussion of
the schemes selected for comparison with our proposed
protocols.
3.1. LEACH. e energy available to the sensor nodes is
limited; therefore energy minimizing protocols are required
International Journal of Distributed Sensor Networks
to overcome this dilemma. Previously few protocols have
been discussed by authors such as direct communication,
multihop communication, and static clustering technique.
ese methods have few drawbacks that need to be addressed.
Along with this, another main issue faced by the sensor nodes
is the bounded bandwidth available for wireless commu-
nications. Hence a protocol is needed in which bandwidth
requirement is controlled.
LEACH deals with the problems mentioned above. It
reduces the energy consumption of nodes in the transmission
and reception of data. It enhances the overall network lifetime
and makes the nodes die randomly in the network. As a result
it avoids the network partitions and does not leave the areas
unattended, which was the drawback of Direct Transmission
(DT) and Multihop Transmission (MT) protocols.
Sensors generate lots of redundant data and their trans-
mission and reception overburden the network. In LEACH
this problem is avoided with the help of data aggression
or fusion technique. rough this way multiple numbers of
unreliable data packets are composed into a single reliable
data packet and as a result communication process is con-
trolled in the overall network.
In LEACH the sensors form a cluster together and elect
a CH aer denite time intervals, randomly depending
upon the residual energies of nodes. LEACH allows every
node to become the CH with equal probability. Aer the
completion of this startup process the non-CH nodes will
decide which cluster they will join. is is decided on the basis
of the Received Signal Strength of the advertisement message
broadcasted by the CHs. Hence, the overall network breaks
up into number of uneven clusters. All the non-CH nodes
will transmit their data destined to that CH. Aerwards
CH aggregates that data and forwards it towards the sink
node. is way the overall transmission distance of the nodes
reduces while compromising the energies of very few nodes
that need to communicate over the large distances. CDMA
codes are used in LEACH to avoid the interference of signals
among the intercluster and intracluster nodes.
Simulationsshowverygoodresultsascomparedtothe
DT and MT transmission techniques. Energy dissipation is
highly reduced in the protocol resulting in larger network
lifetime. In LEACH the death of rst node occurs times
later than the compared protocols. In Direct Transmission
technique the nodes farther to the sink die earlier due to
larger transmission distance while in MT the nodes nearer
to the sink die earlier due to relaying maximum number of
data packets, hence creating the partitions in the network.
Simulations and experimentations show that the nodes in the
LEACH die randomly in the network, which avoids energy
holes in the networks.
Considering all of this, LEACH is a protocol that per-
forms better than the other three techniques while balancing
the energy consumption of the network.
3.2. DEEC. In WSN, data transmission mechanism of sensor
nodes contributes to network lifetime. However, direct and
multihop data transmission techniques failed to achieve
maximum network lifetime. ough, clustering mechanism
proved to be eective for data gathering in WSNs. However,
selection criteria and the number of CHs should be optimum
forprolongationofnetworklifetime.Mostoftheclustering
protocols proposed for WSNs like LEACH, PEGASIS, and
HEED consider homogeneous sensor networks. However, the
residual energy of nodes varies with time and the network will
become heterogeneous.
e CH selection mechanism in protocols like LEACH
failed to achieve prolonged network lifetime under heteroge-
neous network scenario. e node with minimum residual
energy can be selected as CH as all the nodes have same
probabilitytobecomeCHduringnetworklifetime.SEP
considers two-level heterogeneity; however it is required to
nd the optimal probability for CH selection considering the
initial and residual energy of sensor nodes under multilevel
heterogeneous networks.
Tosolvetheaboveissue,DEECalgorithmforhetero-
geneousnetworkhasbeenproposedinthiswork.DEEC
considers multilevel heterogeneous networks by assigning the
initial energy to sensor nodes between 0and 0(1+max),
where 0is the minimum, while max is the maximum energy
value. Also, DEEC assigns rotating epoch 𝑖(probability of
anodetobeCHforcertainnumberofrounds)tosensor
nodes according to their initial and residual energy level. Let
optimal probability opt =1/𝑖,where𝑖is the number of
rounds.Tondtheoptimalandaverageprobability(𝑖)for
CH selection, average energy of network is calculated as
()=1
total 1−
, ()
where is the total rounds during network lifetime. ()is
used as a reference energy value, which is compared against
theresidualenergyofeachnode.So,everynodewillconsume
the same amount of energy during each round. 𝑖should be
dierent than opt (optimal probability) for heterogeneous
nodes. Using ()to be the reference energy,
𝑖=opt 𝑖()
(). ()
So, the optimal number of CHs per round per epoch
is opt . Each node 𝑖determines itself to be a CH using
probability threshold as
𝑖= 𝑖
1−𝑖mod 1/𝑖,()
if 𝑖∈.is the set of eligible nodes to become CH at round
. An eligible node selects a random number between 0 and
1 and compares it against threshold; if the generated number
is less than the threshold, the node becomes CH for current
round. To nd the number of rounds (𝑖)for the node to
remain CH
𝑖=opt ()
𝑖().()
So, the high residual energy nodes will be selected as
CH more oen than the low energy nodes. For multilevel
heterogeneous nodes 𝑖is calculated as
𝑖=opt(1+)𝑖()
((+1)()) ;()
International Journal of Distributed Sensor Networks
and the rotating epoch is
𝑖=()
opt1+𝑖.()
DEEC uses rst order radio model for calculating trans-
mission and reception energy of sensor nodes. According
to this model the total energy consumption in a round is
calculated as
round
=elec +EDA +4BS +2CH, ()
where is the number of bits in a packet, is the number of
clusters, EDA is the energy required for data aggregation at
CH, and BS and CH are average distance of node from
BS and CH, respectively.
In DEEC, BS broadcasts total energy of network along
with the estimated value for the total number of rounds
during network lifetime to all nodes. is information is
needed by every node to calculate 𝑖during each round.
Also using 𝑖node calculates its threshold value and decides
whether it can act as a CH during current round.
Performance of DEEC is evaluated and compared against
LEACH, LEACH-E, and SEP protocols using MATLAB
simulator with -node network. Simulation results showed
that DEEC performs well under multilevel heterogeneous
networks with improved network lifetime and stability period
as opposed to LEACH, LEACH-E, and SEP.
4. Proposed Protocols: HEER and MHEER
Since this research work focused on the improvement of
networks energy eciency and reactive protocols are more
energy ecient than the proactive ones, thereby we have
proposed reactive protocols. In this section, we explain
our proposed protocols HEER [] and MHEER. A number
of routing protocols have been proposed in the eld of
WSNs.Mostoftheminvolveclustering.However,notmuch
attention has been devoted towards time critical applications.
DEEC, being a proactive heterogeneous network protocol,
is not well suited for time critical applications. TEEN is a
reactiveprotocolanditguaranteesthattheunstableregion
would be short in a homogenous network. is is due to
the well-distributed uniform energy consumption in TEEN.
On the other hand, TEEN yields a large unstable region in
a heterogeneous network because the CH selection process
becomes unstable and the nodes stay in idle state for most
of the time. HEER chooses CHs on the basis of residual
energies of the nodes. Because of its reactive nature, it reduces
the number of transmissions and results in better network
lifetime and stability region than TEEN and DEEC. On the
other hand, MHEER yields better results in terms of lifetime
and stability period as compared to HEER. e number of
CHs in HEER is not xed in every round, whereas MHEER
uses static clustering. It also takes into account the maximum
energy nodes at the start of each round for the CH selection.
We explain both proposed protocols in detail in the following
sections.
4.1. HEER. As we have already explained, proactive protocols
sense their environment and transmit data periodically. ey
consume energy continuously due to periodic transmissions.
Main focus in proactive protocols is on increasing lifetime
and throughput and on decreasing energy consumption. In
reactive protocols, a node senses the environment period-
ically but transmits data only when its value reaches the
thresholdvalueoftheattribute.istechniquereducesthe
number of transmissions. Reactive protocols are application
dependent. Keeping in view the fact that data transmission
consumes more energy than data sensing, throughput can be
minimized or maximized as per application of the network.
e throughput in reactive networks is inversely proportional
to the network lifetime or its stability period. So, if the
number of transmissions is less, it will result in extended
stability period as well as network lifetime. However, if the
current Sensed Value reaches the threshold value (absolute
value) repeatedly then maximum number of transmissions
will occur and nodes will die quickly.
In this section, we propose HEER, which improves the
stable region for clustering hierarchy process for a reactive
network in homogeneous and heterogeneous environment.
Similar to DEEC, this protocol also takes into account the
initial and residual energy of nodes for CH selection. When
cluster formation is nished, the CH transmits two threshold
values, that is, Hard reshold (HT) and So reshold
(ST). e nodes sense their environment repeatedly and if
a parameter from the attributes’ set reaches its HT value,
thenodeswitchesonitstransmitterandtransmitsdata.e
Current Value (CV), on which rst transmission occurs, is
stored in an internal variable in the node called Sensed Value
(SV). Now the nodes will again transmit the data to their
respective CHs if
CV −SV ≥ST.()
If the CV diers from SV by an amount equal to or
greater than ST, only then the nodes will transmit their
data. It results in reduced number of transmissions. Figure
shows dierent states of a cluster. e outermost circle in
all the states is referred to as a cluster. Nodes sense their
environment continuously until the parameter (CV) reaches
its HT value. As CV reaches HT value, the nodes start sending
their data to the CH as shown in state (). e CH receives,
aggregates, and then transmits this data to the BS. e CV
on which rst transmission occurs is stored in SV. e node
then again starts sensing its environment as shown in state ()
until the CV diers from SV by an amount equal to or greater
than ST. When this condition is again satised, the node again
switches on its transmitter and sends data to the CH. is data
is then transmitted to the BS by the CH as shown in state ().
4.2. MHEER. An ecient routing protocol is the one []
which consumes minimum energy and also provides good
coverage area. Minimum consumption of energy leads
towards better network lifetime and particularly the stability
period, whereas good coverage area is useful in getting the
required information from the whole network area. e
unattended areas are referred to as coverage holes. ese
International Journal of Distributed Sensor Networks
Data transmissionData transmission
If CV ≥HT
If CV −SV ≥ST
State (1)
State (4)
State (2)
State (3)
Let CV =SV
Normal nodes
High energy nodes
Cluster head
Transmitting data
Base station
F : Idea gure for HEER from data sensing to data transmis-
sion for a cluster.
coverage holes result in inecient coverage area and those
areas can not be monitored. So, the primary objective of a
routing protocol is to achieve minimum energy utilization
and full coverage area. Many researches have addressed such
matters as in [–]. Dierent approaches can be used to
solve this problem, one of which is the division of the network
eld area into subareas. In the proposed technique, we divide
the network area into subareas as explained in the following
subsection.
We consider a WSN of area m × m. e whole area
is divided into ten regions of equal area. Each one of these
regions acts like a cluster. e total number of nodes is .
Each of the regions contains nodes randomly deployed
init.isdivisionhelpstoimprovethecoverageareaofthe
network and all areas are eciently monitored. e network
topologycanbeobservedinFigure .
e network area is divided into equal regions (i.e.,
R–R) as shown in Figure . MHEER uses static clustering.
Static clustering refers to the type of clustering in which clus-
ters are predetermined and they do not change their number
andsizeduringanyround.OnlyoneCHischosenfromeach
region during every round. ese CHs are responsible for the
transmission of data to the BS. All nodes sense their data
and send it to the CH of their region. e CH receives that
data, aggregates it, and then transmitsit to the BS. e energy
consumed during data transmission depends signicantly on
the distance between the CH and the BS. e greater the
distance is, the greater the energy required to transmit that
data to the BS is. All CHs have their own distances from
the BS which depend upon their region and their location
in that region. A CH which is farther from the BS consumes
more energy than the CH which is near the BS. MHEER uses
multihoping technique to cope with this issue. According to
100
75
50
25
0
R1
R2
R3
R4
R6
R5R7
R8
R10
R9
020 40 60 80 100
Normal node Cluster head
Sink location R1–R10 Regions
F : MHEER network topology.
this technique, the CHs which are farther from the BS do not
send their data directly to the BS. Instead, they rst send it
to the CH which is nearer to them as compared to the BS.
ose CHs then forward that data to the BS. According to
Figure , CHs in regions R, R, R, and R do not send
their data directly to the BS. ey calculate their distance
from the CHs of the adjacent regions and then send their
data to the nearer one. In this way, a CH in region R rst
calculates its distance from the CHs of regions R and R and
then transmits its data to the BS via the CH which is near
to it. Similarly, R calculates its distance from R and R,
whereas R and R calculate their distances from R and R.
is multihoping helps to improve the energy consumption
eciency and improves network lifetime and particularly the
stability region.
As in any real case scenario, the number of packets
received at the BS is never equal to the number of packets sent
to the BS. is is because some packets are lost due to certain
factors. ose factors may include interference, attenuation,
andnoise.atiswhyweimplementtheUniformRandom
Distribution Model [] for the calculation of packets drop.
is makes MHEER more practical.
MHEER selects a node as the CH of its region if it
has the maximum energy before the start of that round.
Initially, all nodes have the same amount of energy and any
node can become the CH for rst round. A node is chosen
randomly to become the CH of that region for the rst round.
All other nodes send their data to CH which receives that
data, aggregates it, and sends it to the BS. When the rst
round is completed, the amount of energy in each node
is not the same anymore. is is because the utilization of
energy depends upon the distance between the node or CH
which is transmitting and the CH or sink which is receiving.
Distance is directly proportional to the energy consumption
International Journal of Distributed Sensor Networks
cost of a transmitting node. As distance for transmission
and reception is dierent for dierent nodes, their energy
consumption will also be dierent. For every next round, the
CHs are selected on the basis of maximum energies. e node
with the maximum energy in a region becomes the CH of that
region for that particular round.
5. Sink Mobility in HEER and MHEER:
HEER-SM and MHEER-SM
In this section, we propose the application of sink mobility
on HEER and MHEER and refer to them as HEER-SM and
MHEER-SM, respectively. Sink mobility has been proved
very eective in extending the network lifetime and particu-
larly the stability region. We put greater emphasis on stability
region because this is the region in which the data received
attheBSismostreliableaseverynodeisaliveduringthis
region. So, in terms of data integrity, stability region is very
important. Multiple mobile sinks would signicantly prolong
thenetworklifetimeandmaximizethethroughput.However,
the installation cost would also signicantly increase. us,
to prolong the network lifetime and maximize throughput
while keeping the installation cost within a fairer limit, we
have used only one mobile sink.
Sink mobility refers to the movement of sink in the
network to collect the data from the static nodes. ese nodes
can be either normal nodes or CHs, depending upon its
application. Sink mobility is of two types, controlled and
uncontrolled mobility. For the latter, the mobile sink can
move randomly in the network region, whereas for the former
itcanonlymovealongthepredenedtrajectory.Controlled
mobility can be implemented by two ways. In the rst way, the
sinkcanmoveinthenetworkonitspredenedlocationsand
thesepredenedlocationscannotbechangedthroughoutthe
network lifetime. While according to the second way the sink
moves on its predened locations these locations are changed
aer every round. In this way, the sink moves in the controlled
fashion, but its trajectory is changed aer every round. In
ourtechnique,weimplementtheformermethodinwhich
thesinklocationsarepredenedandtheyarenotchanged
throughout the network lifetime. ese sink locations are
also referred to as sojourn locations. e sink stops at these
locations to collect the data from the nodes/CHs.
5.1. Network Topology. e number of sinks is restricted to
one. All nodes in the network are static; that is, they do
not move. e sink moves between dierent regions in the
network area under consideration. It stops at certain sojourn
locations and collects the data from the nodes. In order to
minimize the communication distance between nodes of a
givensubregionandmobilesink,thesojournlocationsare
chosen as the centre points of each subregion. Figure shows
the network topology of our proposed sink mobility. e ×
marks show the sojourn locations in the network. e sink
ismountedonanunmannedremotecontrolledvehicleand
moves from one sojourn location to the next and collects data
from the nodes at these sojourn locations. e nodes collect
dataandsendittotheirrespectiveCHs.emobilesinkstops
at its sink stops and collects data from nodes or CHs.
100
75
50
25
0
R1
R2
R3
R4
R6
R5R7
R8
R10
R9
0 20 40 60 80 100
Normal node Cluster head
Sink location R1–R10 Regions
F : Sink mobility.
e whole travel distance covered by the sink in the whole
network lifetime should be bounded because a mobile sink
is usually driven by fuel or electricity. When a mobile sink
moves from one sink location to another, probability of data
loss is high, so the distance between two sink locations should
be restricted. e transmission of data from nodes/CHs to
sink only occurs when the sink is not moving; that is, sink is
located at any sink location. erefore, the sum of stop times
inthemobilesinktourshouldbemaximized.ereshouldbe
maximum number of stop locations of mobile sink, as could
be seen from Figure .
5.2. Clustering Mechanism. In our model, a single sink moves
around the network to collect the data from the nodes/CHs
from its sink locations. ese sink locations are predened
and do not change throughout the network lifetime.
In MHEER, the sink moves to each region and stops at its
specic sojourn location to collect that data. As there are
regions, and the sink has to collect the data from all regions,
there are sink stops predened. ese stops are located
in the middle of each region. e CHs collect data from the
nodes, aggregate it, and send it to the sink whenever the sink
comes to their region to collect the data. In case of HEER, the
area is not divided into subregions. But the sink locations for
HEER are also the same as that for MHEER. e dierence
is that in MHEER each region is predened and each region
has its own CH to transmit the data to the sink. So, when the
sink arrives, the CH sends its data to it, whereas, in HEER,
the regions are not predened and clusters change their shape
and size. e number of CHs is not the same. So, each CH
calculates its distance from its neighbouring sink locations
andassociatesitselfwiththeclosestone.enormalnodes,
in addition to calculating their distance from the CH, also
calculate their distance from the sink location. ese nodes
International Journal of Distributed Sensor Networks
Start
Parameter
initialization
Formation of
regions
Nodes
deployment
First
round?
Select any
node as CH Ye s
Ye s
Ye s
Network
average
energy
calculation
No No
No
No
No
Maximum
energy node?
Select as
normal node
Select as CH
Sense CV If CV ≥H
SV =CV
Ye s
Ye s
Ye s
If CV −SV ≥S
If residual
energy of all
nodes >0
Network
dead
End
No
Is it a CH
node?
Data
transmission
from CH to BS
Data
transmission
from normal
node to CH
F : Flowchart of MHEER protocol.
T : Simulation parameters.
Parameter Value
Number of nodes , ,
Number of static sinks
Number of mobile sinks
Static sink location (, )
Mobile sink location Specied trajectory (reference Figure )
Initial energy . J
Area m × m
then send their data to the one which is closer to them than
the others. In this way, energy is quite eciently consumed.
6. Experiments and Discussions
In this section, we discuss the simulation results of our pro-
posed protocols. Tab l e summarizes the simulation parame-
ters used to validate the proposed protocols.
6.1. Performance Metrics: Denitions. e following perfor-
mance metrics are considered:
() Network lifetime: It is the time period from the start
of the network till the death of the last node in the
eld. It is measured in the unit of time (seconds). It is
one of the most important parameters every network
is supposed to have.
() roughput: It is the total number of packets success-
fullyreceivedattheBS.Itexcludesthepacketsentby
the sensor nodes in the eld but dropped on their way
to BS because of any reason. Its unit is packets/sec.
() Packet drop: It is dened as the number of packets
sent towards the BS; however, they are not received
at BS.
() Total energy consumption: It is dened as the total
energy consumed by all the alive nodes. It is measured
in Joules.
() End-to-end delay: It is the total time taken by all
packets to reach from source node to BS. It is also
measured in seconds.
6.2. Performance Metrics: Discussions. In this section, we
discuss the performance parameters by which we measure,
evaluate, and then compare our proposed protocols with
the existing counterpart protocols. For the sake of fair
comparison, we have assumed the So and Hard reshold
ranges as in the selected protocol for comparison, that is,
TEEN. Similar reason is valid for initial energy of nodes.
6.2.1. Network Lifetime. To understand the network lifetime,
we rst dene the alive nodes. e nodes with sucient
energy to sense, process, and then transmit the data to the
neighbors, and/or BS or any other node in its transmission
range, are generally referred to as alive nodes. Generally,
the lifetime of any network is depending upon the number
of alive nodes (which in fact is depending upon the initial
energy and consumption of energy). As per our assumption,
International Journal of Distributed Sensor Networks
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
0
20
40
60
80
100
Time (s)
Alive nodes
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(a) Lifetime of network for nodes
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
0 0.05 0.1 0.15 0.2 0.25
0
100
200
300
400
500
Time (s)
Alive nodes
(b) Lifetime of network for nodes
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
0 0.1 0.2 0.3 0.4 0.5 0.6
0
200
400
600
800
1000
Time (s)
Alive nodes
(c) Lifetime of network for nodes
F : Lifetime of network for dierent numbers of nodes.
even if a single alive node in the network is working, the
network is assumed alive. High energy consumption could
result in short lifetime and vice versa. e ecient routing
protocols generally result in the ecient consumption of
energy which ultimately improves the network lifetime.
In Figure , we compare the network lifetime of TEEN,
DEEC, HEER, MHEER, HEER-SM, and MHEER-SM. We
can see that MHEER-SM has the best lifetime as compared
to the other protocols, whereas TEEN has the least network
lifetime.isisbecauseMHEER-SMhasthesamenetwork
topology as that of MHEER with the exception that MHEER-
SM has mobile sink. is mobile sink moves to every region
and collects data from the CH of each region. In this way, the
distance between the CHs and sink reduces, which results in
ecient consumption of energy. We can observe that HEER
outperforms TEEN and DEEC. is is because HEER selects
the CHs on the basis of their residual energies. e data
is transmitted only when the threshold limit is achieved. It
further reduces the number of transmissions and improves
the network lifetime. MHEER on the other hand outperforms
HEER. is is because MHEER is based on multihoping
and the distant CHs transmit their data via multihoping.
In this way, the energy is eciently consumed. MHEER
has static clusters and each cluster has one CH and xed
a number of nodes. is helps in improving coverage area
and coverage holes are reduced. HEER-SM and MHEER-SM
perform better than HEER and MHEER because mobility
helps to reduce the distance between the CHs and the sink. In
this way, the network lifetime and stability region are further
improved. Lifetime and nodes dying at frequent intervals are
given in Tab l e .
6.2.2. Lifetime Maximization Model. Our proposed protocol
modelsaWSNasagraph={∪
0,∪0},where
and 0are the set of sensors and sink locations, respectively.
=||isthenumberofsensorsand0=|
0|is the
number of sink sites. ={∪}is the set of wireless links
between sensors nodes and 0={∪0}is the set of wireless
links between sensor nodes and sink locations. 𝑖𝑐 ∈if the
CH is within the communication range 𝑖of node ,where
∀, ∈ .Similarly,𝑖𝑘 ∈
0if the sink location is within
the communication range of node ,and𝑐𝑘 ∈0if the CH
is within the communication range of sink location ,where
∀∈0.𝑖is the data generation rate of a node and its value
isthesameforallnodes.esinkspeedistakenasinnity.
e sink has unlimited energy and there is no energy
issue for sink. e residual time of the sink at each location
is dened as 𝑘. Nodes send their data only during this time.
Nodes do not send their data whenever the sink is in motion.
𝑖𝑐 is the amount of data from node to CH .𝑖𝑘 is the
amount of data from node to the sink location .And𝑐𝑘 is
the amount of data from CH to the sink location .=1, if
the sink is at the sink site of a region. 𝑇
𝑖𝑐 is the energy required
totransmitoneunitdatafromnodeto CH .eenergy
dissipated for the transmission of one unit data from node
International Journal of Distributed Sensor Networks
T : Dead nodes at dierent instants of time (for nodes).
Protocol name First node dead at Last node dead at Dead nodes
. sec . sec . sec . sec
HEER . sec . sec
HEER-SM . sec . sec
MHEER . sec . sec
MHEER-SM . sec . sec
TEEN . sec . sec
DEEC . sec . sec
to sink location is dened as 𝑇
𝑖𝑘. e amount of energy
consumed for the reception of one unit data is given as 𝑅
𝑐𝑖.
Since the objective function and its given constraints
are mixed integer nonlinear, we have chosen mixed integer
nonlinear programming model:
Maximize =
𝑟
𝑘𝑟
𝑘()
Subject to:
𝑟𝑇
𝑐𝑘
ℓ𝑐𝑘∈𝜐0𝑟
𝑐𝑘 +𝑅
𝑗𝑐
ℓ𝑗𝑐∈𝜐 𝑟
𝑗𝑐≤𝑐,
𝑟
𝑖=1, ∀,∈, ∀∈0,(9a)
𝑟𝑇
𝑖𝑘
ℓ𝑖𝑘∈𝜐0𝑟
𝑖𝑘 +𝑇
𝑖𝑐
ℓ𝑖𝑐∈𝜐 𝑟
𝑖𝑐≤𝑖,
𝑟
𝑖=0, ∀,∈, ∀∈0,(9b)
𝑖𝑖𝑐 =0,
i 𝑖𝑐 ≤𝑖𝑘, ∀,∈, ∀∈0,(9c)
𝑘>0,
𝑖𝑗 ≥0, ∀,,, (9d)
∀,,∈, ∀∈0.(9e)
is model is a mixed integer nonlinear programming model.
We explain each equation below.
Objective Function. e objective function of this sink mobil-
ity model is to maximize the sojourn time of sink. e reason
behind it is that the sink collects the data from the nodes or
CHsonlywhenitisatitssinklocation.Itdoesnotcollect
thedatawhenitisinmotion.So,aslongasthesinkstays
at its sink location, it collects the data. In this way, improving
the sojourn time will result in improving the network lifetime
which is our main goal.
Energy Constraint. According to these constraints, if a node is
a CH, it receives the data from the nodes and sends that data
to the sink. e energy consumed during this process should
be less than that of the initial energy of the CH. Similarly, if
a node is not a CH, then it will either send its data to the CH
or to the sink depending upon their distance from that node.
is consumption of energy should also be less t han the initial
energy of the node.
Flow Constraint. is constraint shows that a node only sends
its data to the CH if and only if the distance of the CH from
the node is lesser than the distance between the sink and the
node. If this distance is greater, then the node transmits its
data directly to the sink instead of sending its data to the sink
via the CH.
PauseTimeConstraint. is constraint says that every pause
time of the sink must be greater than zero because sink does
not collect data when it is in motion. It collects the data of
the nodes or the CHs only when it is at its sink site. So, the
sojourntimeshouldbegreaterthanzerotocollectthatdata.
6.2.3. roughput. In this subsection, we discuss the number
of packets sent to the BS. Figure shows the total number of
packets sent to BS, in TEEN, DEEC, HEER, MHEER, HEER-
SM, and MHEER-SM. We know that the CH selection in
HEER and HEER-SM is based on the probability assigned to
each node. is results in uneven number of CHs in each time
round. As the number of CHs in each time round is not the
same, the number of packets sent to the BS per time round is
also not xed. e number of packets sent to the BS varies in
every time round. As the probability of CHs per time round in
HEER and HEER-SM is ., the number of CHs in every time
roundshouldbe.SothenumberofpacketssenttotheBS
should also be . But the number of CHs does not remain
xed. As a result, the number of packets sent to BS is also not
the same.
In case of MHEER and MHEER-SM, the selection of CHs
is based on the maximum residual energy of a node in its
region. As there are regions and every region has CH in
each time round, the number of CHs in each time round is
also . Each CH is responsible for sending its data to the BS.
So,packetssenttotheBSineverytimeroundarealso.
6.2.4. Data Gathering Maximization Model. We also dene a
new model for the data gathering. In this model, we maximize
the data gathering at the sink which results in maximized
throughput. Maximum throughput leads to the conclusion
that maximum data is gathered at the sink. is total data
gathering 𝑘at the sink site can be dened as the sum of
the data transmitted by the nodes and the CHs to the sink
site. min is the minimum sojourn time for which sink stays at
International Journal of Distributed Sensor Networks
0 0.01 0.02 0.03 0.04 0.05
0
0.5
1
1.5
2
2.5
Time (s)
Packets to BS
×105
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(a) roughput of network for nodes
0 0.05 0.1 0.15 0.2 0.25
0
2
4
6
8
10
12
14
Time (s)
Packets to BS
×105
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(b) roughput of network for nodes
0 0.1 0.2 0.3 0.4 0.5 0.6
0
0.5
1
1.5
2
2.5
Time (s)
Packets to BS
×106
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(c) roughput of network for nodes
F : Packets sent to BS.
site .ℵis the total number of regions. It can be given by the
following equation:
𝑘=
𝑟
ℓ𝑖𝑘∈𝜐0𝑟
𝑖𝑘 +
ℓ𝑐𝑘∈𝜐0𝑟
𝑐𝑘, ∀,∈, ∀∈0.()
is is also a mixed integer nonlinear programming model:
Maximize 𝑘, ()
subject to: 𝑟
𝑘≥min ,∀∈
0,∀, (11a)
0=lim ×
ℵ,where max ≥≥1,(11b)
0=lim ×
ℵ,where max ≥≥1.(11c)
Objective Function. We maximize the data gathering at the
sink which results in maximized throughput. Maximum
throughput leads to the conclusion that maximum data is
gathered at the sink. is total data gathering 𝑘at the sink
site can be dened as the sum of the data transmitted by the
nodes and the CHs to the sink site.
Sojourn Time Constraint. According to this constraint,
increasing the sojourn time increases the amount of data
gathering. is is because the greater the time the sink stays
at its sink location, the greater the time nodes and CHs get
tosendtheirdatatothesink.So,asinkshouldstayformore
time at its sink location than its least possible sojourn limit.
is results in maximum data gathering.
Sink Locations Constraint. is constraint discusses the num-
ber of sink sites. e greater the number of sink sites, the
greatertheamountofdatagathered.Soifasinkstopsatmore
locations, it gathers more data than the sink which stays at few
locations. According to this constraint, the number of sink
locations should be according to the network area and the
number of regions in it. So, the number of sink sites can be
determined by using the maximum limits of network’s length
,width,andthenumberofregions.
6.2.5. Packet Drop. Packet drop can be dened as number of
total packets sent minus the total number of packets received.
Interow and intraow interferences, congestion, path loss,
attenuation, noise, and so forth could be the reasons for
thepacketdrops.Inourproposedtechniques,wehaveused
the uniform random distribution to calculate the number
of dropped packets. We assume that it makes our protocol
relatively robust as compared to the counterpart schemes. We
use . as the packet drop probability value, which means,
during every time round, a packet has a probability of %
to be dropped.
International Journal of Distributed Sensor Networks
0 0.01 0.02 0.03 0.04 0.05
0
0.5
1
1.5
2
2.5
Time (s)
Packets received
×104
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(a) Total number of packets dropped for nodes
0 0.05 0.1 0.15 0.2 0.25
0
2
4
6
8
10
12
Time (s)
Packets received
×104
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(b) Total number of packets dropped for nodes
0 0.1 0.2 0.3 0.4 0.5 0.6
0
0.5
1
1.5
2
2.5
Time (s)
Packets received
×105
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(c) Total number of packets dropped for nodes
F : Packet drop for whole network lifetime for , , and nodes.
Figures and show the total number of packets trans-
mittedinthenetworkandnumberofpacketssuccessively
received only at the BS, respectively. From the gures it can
be observed that total number of packets received at the BS is
remarkably less than the total number of packets sent in the
whole network.
6.2.6. Energy Consumption. In this section, total energy
consumption analysis is presented. Total energy includes the
energy required for transmission, reception, and aggregation.
Energy consumption of the network is inversely proportional
to the network lifetime. From Figures and it is obvi-
ous that shorter network lifetime results in greater energy
consumption. Figure compares the energy consumption
of HEER, HEER-SM, MHEER, MHEER-SM, TEEN, and
DEEC. In the beginning, TEEN and DEEC have maximum
energy consumption as compared to remaining protocol
plots. DEEC being proactive protocol consumes more energy
because of periodic transmissions and addition of advanced
nodes which adds to total energy consumption count.
TEEN being reactive protocol consumes more energy
among all reactive protocols because of random selection of
CH and dies out faster. HEER is also a reactive protocol;
however,ittakesintoaccountresidualandinitialenergy
of nodes in CH selection process. erefore, it has better
energy consumption because of load balancing as compared
to TEEN and DEEC. Although stable region of HEER is more
than TEEN and DEEC (Figure ), uctuations in plots of
HEER are due to its reactive nature. Protocol may have few
or many transmissions in any particular time round. e CH
selection is dynamic in HEER; therefore, it has more energy
consumption than MHEER and MHEER-SM. In dynamic
CH selection process sometimes CH is far from BS and more
transmission energy is consumed. MHEER consumes less
energy and shows better network lifetime because of static
clustering and nondistant transmissions from CH to BS. In
case CH is far from BS, it transmits data to nearest CH instead
of sending it to BS.
e introduction of mobile sink in HEER-SM yields less
energy consumption because sink may be more closer than
CH or BS to receive data. Hence, nodes or CHs do not do
distant transmissions. However, sink mobility in MHEER
hasverylittleimpactonenergyconsumptionandnetwork
lifetime because the static clustering architecture of MHEER
is enough to achieve such lifetime with the current mobility
pattern of sink.
6.2.7. Delay. Figure shows the end-to-end delay of the
network.isdelayincludesthetimerequiredbyallthealive
nodes to transmit data to CH and from CH to BS.
From Figure , it can be seen that sink mobility improves
delay performance. MHEER-SM has .% less delay as
International Journal of Distributed Sensor Networks
0 0.01 0.02 0.03 0.04 0.05
0
0.01
0.02
0.03
0.04
0.05
Time (s)
Energy consumption (J)
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(a) Energy consumption for nodes
0 0.05 0.1 0.15 0.2 0.25
0
0.05
0.1
0.15
0.2
0.25
Time (s)
Energy consumption (J)
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(b) Energy consumption for nodes
0 0.1 0.2 0.3 0.4 0.5 0.6
0
0.1
0.2
0.3
0.4
0.5
Time (s)
Energy consumption (J)
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(c) Energy consumption for nodes
F : Energy consumption of the network for dierent numbers of nodes.
compared to MHEER and results are even better in HEER-SM
with % less delay in comparison to HEER. e improve-
ment in delay performance is because of availability of sink
in close vicinity aer frequent intervals. Nodes, instead of
transmitting data to CH and then CH taking another few
seconds to transmit data to BS, transmit directly to mobile
sink. MHEER chooses maximum energy node as a CH which
thentransmitsdatatoBS.CHcanbeatanylocationin
the particular region and may not be nearest to all the
cluster members in that region. However, in MHEER-SM the
sojourn locations of MS are almost in the centre of every
region that makes it feasible for all the cluster members and
also CHs to transmit data with minimum delay and energy
when mobile sink is there.
Delay dierence in HEER and DEEC is small in the
beginning because both are following the same CH selection
criteria; however, it increases later on because of dierence in
lifetime of both protocols. HEER has many alive nodes when
DEEC dies out completely (ref. Figure ).
Another observation from Figure is that static cluster-
ing protocols like MHEER and MHEER-SM have less delay
as compared to dynamic clustering protocols like TEEN,
DEEC, HEER, and HEER-SM. e location of CHs in case
of dynamic clustering is not xed. CH and cluster members
maybetoocloseortoofarfromBSandCH,respectively.In
addition, number of CHs is not xed. erefore, there may
be less than optimal number of CHs in any particular time
round that leads to unbalanced regions. Nodes and CH as
a result do too many distant transmissions and thus add more
delay.
6.3. Performance Trade-Os Made by HEER/MHEER
and HEER-SM/MHEER-SM. In our scheme of MHEER,
improvement in end-to-end delay is achieved at the cost
of frequent transmissions due to packet loss, whereas in
MHEER-SM the end-to-end delay is achieved at the cost
of greater energy consumption, as shown in Table .e
end-to-end delay of the network in MHEER and MHEER-
SM is improved compared to HEER and HEER-SM. e
improvement in delay performance is because of availability
of sink in close vicinity aer frequent intervals. Instead
of transmitting data to cluster head and then cluster head
forwarding data to base station aer some time delay, nodes
transmit data directly to mobile sink. In MHEER-SM, the
sojourn locations of mobile station are almost in the center of
every region that makes it feasible for all the cluster members
and also cluster heads to transmit data with minimum delay
when mobile sink is there. Delay dierence in HEER and
DEEC is small in the beginning because both are following
the same cluster head selection criteria; however, it increases
lateronbecauseofdierenceinlifetimeofbothprotocols.
In MHEER and MHEER-SM, the stability period is
improved because of the same network topology in both
schemes with the exception that MHEER-SM has mobile
sink. is mobile sink moves to every region and collects data
from the CH of each region. In this way, the distance between
International Journal of Distributed Sensor Networks
0 0.01 0.02 0.03 0.04 0.05
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Time (s)
End-to-end delay (s)
×10−5
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(a) End-to-end delay of the network for nodes
0 0.05 0.1 0.15 0.2 0.25
0
1
2
3
4
5
6
7
Time (s)
End-to-end delay (s)
×10−5
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(b) End-to-end delay of the network for nodes
0 0.1 0.2 0.3 0.4 0.5 0.6
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Time (s)
End-to-end delay (s)
×10−4
HEER-SM
HEER
MHEER
MHEER-SM
TEEN
DEEC
(c) End-to-end delay of the network for nodes
F : End-to-end delay of the network for , , and nodes.
T : Performance trade-os made by the protocols.
Protocol Achieved parameter Figure Compromised parameter Figure
HEER End-to-end delay improves Figure roughput Figure
MHEER End-to-end delay improves Figure Frequent transmissions due to packet loss Figure
HEER-SM End-to-end delay improves Figure Energy consumption Figure
MHEER-SM End-to-end delay improves Figure Energy consumption Figure
HEER Stability period extends Figure End-to-end delay Figure
MHEER Stability period extends Figure Redundant transmissions due to packet loss Figure
HEER-SM Stability period extends Figure Greater energy consumption Figure
MHEER-SM Stability period extends Figure roughput Figure
HEER Lifetime extends Figure End-to-end delay and energy consumption Figures and
MHEER Lifetime extends Figure roughput Figure
HEER-SM Lifetime extends Figure End-to-end delay Figure
MHEER-SM Lifetime extends Figure roughput Figure
the CHs and sink reduces, which results in ecient energy
consumption. e stability period of MHEER is improved
butatthecostofredundanttransmissionsduetopacketloss
at the sink. e stability period of MHEER-SM is improved
but at the cost of the network throughput. MHEER has static
clusters and each cluster has one CH and xed a number of
nodes. is helps in improving coverage area and coverage
holes are reduced.
HEER-SM and MHEER-SM perform better than HEER
and MHEER because mobility helps to reduce the distance
between the CHs and the sink. In this way, the network life-
time and stability region are further improved. Fluctuations
International Journal of Distributed Sensor Networks
in plots of HEER are due to its reactive nature. Protocol may
have few or many transmissions in any particular time round.
Cluster head selection is dynamic in HEER; therefore it has
more energy consumption than MHEER and MHEER-SM. In
dynamic selection process, CH may be far from BS and more
transmission energy is consumed. MHEER consumes less
energy and shows better network lifetime because of static
clustering and nondistant transmissions from CH to BS. In
case CH is far from BS, it transmits data to nearest CH instead
of sending it to BS.
In MHEER and MHEER-SM, network lifetime is
improved at the cost of net throughput of the network.
e drop in network lifetime in MHEER and MHEER-SM
is much less than that of other schemes. In HEER and
HEER-SM, network lifetime is improved at the cost of end-
to-end delay of the network and energy consumption. HEER
improves the stable region for clustering hierarchy process
for a reactive network in homogeneous and heterogeneous
environment. e nodes sense their environment repeatedly
and if a parameter from the attributes set reaches its HT
value,thenodeswitchesonitstransmitterandtransmits
data.IncaseofHEER,theareaisnotdividedintosubregions.
But the sink locations for HEER are also the same as that
for MHEER. e dierence is that in MHEER each region
is predened and each region has its own CH to transmit
the data to the sink, whereas, in HEER, the regions are not
predened and clusters change their shape and size.
7. Conclusion
In this paper, we have proposed two scalable routing pro-
tocols, HEER and MHEER. e proposed techniques select
the cluster heads based upon the residual energy of the
nodes. Fixed number of cluster heads (only in MHEER)
areselectedineachcycleofprotocoloperation;wecall
it “round.” Because of HEER’s reactive nature, it reduces
the number of transmissions and results in better network
lifetime and stability region than TEEN and DEEC. On the
other hand, MHEER yields better results in terms of lifetime
and stability period as compared to HEER. e number of
CHs in HEER is not xed in every round, whereas MHEER
uses static clustering. It also takes into account the maximum
energy nodes at the start of each round for the CH selection.
HEER and MHEER are then incorporated with mobile sinks.
As mobile sink has no constraints of energy consumption,
it enhances the working of both techniques. Ultimately,
HEER and MHEER outperform TEEN, DEEC, and HEER
protocols in network lifetime, stability period, area coverage,
and throughput. us, these schemes enhance the desired
attributes, minimum energy consumption, maximum stabil-
ity period, better lifetime, and throughput, as compared with
other protocols. irdly, both of the proposed techniques are
scalable; that is, they even perform well when run for and
nodes in the same scenarios.
Conflict of Interests
e authors declare that there is no conict of interests
regarding the publication of this paper.
References
[] W. R. Heinzelman, A. Chandrakasan, and H. Balakrish-
nan, “Energy-ecient communication protocol for wireless
microsensor networks,” in Proceedings of the 33rd Annual
Hawaii International Conference on System Sciences (HIC SS ’00),
pp. –, IEEE, January .
[] A. Manjeshwar and D. P. Agrawal, “TEEN: a routing protocol for
enhanced eciency in wireless sensor networks,” in Proceedings
of the 15th International Symposium on Parallel and Distributed
Processing, pp. –, San Francisco, Calif, USA, April
.
[] G. Smaragdakis, I. Matta, and A. Bestavros, “SEP: a stable
election protocol for clustered heterogeneous wireless sensor
networks,” in Proceedings of the 2nd International Workshop on
Sensor and Actor Network Protocols and Applications (SANPA
’04), pp. –,B oston University Computer Science Department,
.
[] L. Qing, Q. Zhu, and M. Wang, “Design of a distributed energy-
ecient clustering algorithm for heterogeneous wireless sensor
networks,” Computer Communications,vol.,no.,pp.–
, .
[]K.Latif,A.Ahmad,N.Javaid,Z.A.Khan,andN.Alrajeh,
“Divide-and-rule scheme for energy ecient routing in wireless
sensor networks,” in Proceedings of the 4th International Confer-
ence on Ambient Systems, Networks and Technologies (ANT ’13),
pp. –, June .
[] S.Priyankara,K.Kinoshita,H.Tode,andK.Murakami,“Aclus-
tering method for wireless sensor networks with heterogeneous
node types,” IEICE Transactions on Communications,vol.,
no. , pp. –, .
[] V.T.QuangandT.Miyoshi,“Adaptiveroutingprotocolwith
energy eciency and event clustering for wireless sensor net-
works,” IEICE Transactions on Communications,vol.,no.,
pp. –, .
[] Y. Kiri, M. Sugano, and M. Murata, “On characteristics of multi-
hop communication in large-scale clustered sensor networks,”
IEICE Transactions on Communications,vol.E-B,no.,pp.
–, .
[] N. Javaid, S. N. Mohammad, K. Latif, U. Qasim, Z. A. Khan, and
M. A. Khan, “HEER: hybrid energy ecient reactive protocol
for wireless sensor networks,” in Proceedings of the 2nd IEEE
Saudi International Electronics, Communications and Photonics
Conference (SIECPC ’13), pp. –, IEEE, Riyadh, KSA, April
.
[] A. Ogawa, T. Yamazato, and T. Ohtsuki, “Information and
signal processing for sensor networks,” IEICE Transactions on
Fundamentals of Electronics, Communications and Computer
Sciences A,vol.,no.,pp.–,.
[] H. Saito and H. Minami, “Performance issues and network
design for sensor networks,” IEICE Transactions on Communi-
cations B,vol.,no.,pp.–,.
[] Y.Gu,Y.Ji,J.Li,andB.Zhao,“ESWC:ecientschedulingfor
the mobile sink in wireless sensor networks with delay con-
straint,” IEEE Transactions on Parallel and Distributed Systems,
vol.,no.,pp.–,.
[]W.Liang,J.Luo,andX.Xu,“Prolongingnetworklifetime
via a controlled mobile sink in wireless sensor networks,” in
Proceedings of the 53rd IEEE Global Communications Conference
(GLOBECOM ’10), pp. –, December .
[] J. Luo and J.-P. Hubaux, “Joint sink mobility and routing to
maximize the lifetime of wireless sensor networks: the case of
International Journal of Distributed Sensor Networks
constrained mobility,” IEEE/ACM Transactions on Networking,
vol.,no.,pp.–,.
[] A. W. Khan, A. H. Abdullah, M. A. Razzaque, and J. I. Bangash,
“VGDRA: a virtual grid-based dynamic routes adjustment
scheme for mobile sink-based wireless sensor networks,” IEEE
Sensors Journal,vol.,no.,pp.–,.
[] Y. Chen, Z. Wang, T. Ren, and H. Lv, “Lifetime optimization
algorithm with mobile sink nodes for wireless sensor networks
based on location information,” International Journal of Dis-
tributed Sensor Networks.Inpress.
[] S. Jabbar, A. A. Minhas, M. Imran, S. Khalid, and K. Saleem,
“Energy ecient strategy for throughput improvement in wire-
less sensor ne tworks,” Sensors,vol.,no.,pp.–,.
[] L. Shi, J. Zhang, Y. Shi, X. Ding, and Z. Wei, “Optimal base
station placement for wireless sensor networks with successive
interference cancellation,” Sensors,vol.,no.,pp.–,
.
[] R.SugiharaandR.K.Gupta,“Sensorlocalizationwithdeter-
ministic accuracy guarantee,” in Proceedings of the IEEE INFO-
COM, pp. –, April .
International Journal of
Aerospace
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Robotics
Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Active and Passive
Electronic Components
Control Science
and Engineering
Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
International Journal of
Rotating
Machinery
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com
Journal of
Engineering
Volume 2014
Submit your manuscripts at
http://www.hindawi.com
VLSI Design
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Shock and Vibration
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Civil Engineering
Advances in
Acoustics and Vibration
Advances in
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Electrical and Computer
Engineering
Journal of
Advances in
OptoElectronics
Hindawi Publishing Corporation
http://www.hindawi.com
Volume 2014
The Scientic
World Journal
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Sensors
Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Modelling &
Simulation
in Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Chemical Engineering
International Journal of
Antennas and
Propagation
International Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Navigation and
Observation
International Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Distributed
Sensor Networks
International Journal of
