IMPACTS OF STRUCTURAL FACTORS ON ENERGY CONSUMPTION IN CLUSTER-BASED WIRELESS SENSOR NETWORKS: A COMPREHENSIVE ANALYSIS

International Journal of Ad hoc, Sensor & Ubiquitous Computing (IJASUC) Vol.6, No.1, February 2015
DOI : 10.5121/ijasuc.2015.6101 1
IMPACTS OF STRUCTURAL FACTORS ON
ENERGY CONSUMPTION IN CLUSTER-BASED
WIRELESS SENSOR NETWORKS: A
COMPREHENSIVE ANALYSIS
Taner Cevik1
and Fatih Ozyurt2
1
Department of Computer Engineering, Fatih University, Istanbul, Turkey
2
Department of Software Engineering, Firat University, Elazig, Turkey
ABSTRACT
Limited energy is the major driving factor for research on wireless sensor networks. Clustering alleviates
this energy shortage problem by reducing data traffic conveyed over the network and therefore several
clustering methods are proposed in the literature. Researchers put forward their methods by making
serious assumptions such as always locating single sink at one side of the topology or making clusters near
to the sink with smaller sizes. However, to the best of our knowledge, there is no comprehensive research
that investigates the effects of various structural alternatives on energy consumption of wireless sensor
networks. In this paper, we thoroughly analyse the impact of various structural approaches such as cluster
size, number of tiers in the topology, node density, position and number of sinks. Extensive simulation
results are provided. The results show that the best performance about lifetime prolongation is achieved by
locating a sufficient number of sinks around the network area.
KEYWORDS
Wireless Sensor Networks, Clustering, Energy Conservation, Network Lifetime.
1. INTRODUCTION
Incredibly small sized sensor nodes have recently become available on the market with affordable
prices facilitating technological improvements in microelectronics, signal processing, etc. which,
in turn allowed application areas of such sensor nodes in our daily lives to expand rapidly [1-2].
These devices constitute mainly three sub-units: the processor, the sensing and the
communication units [3]. Since these nodes can be self-organized without any intervention after
the deployment stage, they form a Wireless Sensor Network (WSN) which is a subclass of ad-hoc
networks [4]. However, there are significant differences between WSNs and their other ad-hoc
counterparts. First, the nodes in traditional ad-hoc networks communicate mostly in a point-to-
point manner. However, since the nodes in WSNs have limited energy sources, they prefer to
communicate in a multi-hop manner. As pointed out by Akyildiz et al. [5], another important
difference is that nodes in WSNs are deployed in a more intensive manner than the nodes
deployed in traditional ad-hoc networks. Therefore, using the protocols and the methods utilized
for ad-hoc networks will not be effective for WSNs.
As well documented in the literature, the most important drawback of these sensor nodes is the
energy expenditure. Thus, in order to use thousands or millions of these devices in a topology,
energy-aware protocols and architectures should be considered [6-8].

2
The major energy consuming unit of a sensor node is the communication unit. Raghunathan et al.
[9], established that sensor nodes consume much more energy during data communication when
compared with data processing. Hence, researchers have dramatically focused on developing
energy-efficient communication protocols and architectures. Most prominent categories are duty-
cycling methods, data-driven approaches, and clustering.
The key point in duty-cycling is defining a sub-tree of nodes in the topology that will remain
awake while the others go to sleep. In this way, communication throughout the network is still
active while only a portion of the nodes stay awake. Another important point is to define suitable
sleep and wake-up schedules for these nodes in order to provide the sustainability of the network.
Two major subcategories constituting the data driven approach are data acquisition and data
aggregation methods which aim to reduce the amount of data to be conveyed. Data acquisition is
performed at signal level. In contrast, data aggregation is performed at application level. Data
acquisition is adding distinct signals and transmitting data as a single aggregate. However, data
aggregation is something like filtering and summarizing the original data coming from all sensor
nodes.
Another very popular category of methods for lifetime prolongation is clustering. The main idea
in clustering is grouping the sensor nodes depending on a number of criteria, in other words,
virtually partitioning the topology into grids. Clustering can provide significant energy savings
especially in high density networks. In 2011, Kumar et al. indicated that, since the plain nodes in
clusters direct their data to their cluster heads, problems often encountered such as multiple
routes, flooding and routing loops are eliminated or alleviated [10].
This paper presents a comprehensive analysis of the effects of the various structural factors in
terms of energy consumption in WSNs. General belief about cluster-based WSNs is that in order
to alleviate the hot-spot problem, clusters located near the sink should be smaller-sized than the
ones further from the sink. Other possible factors that may affect the lifetime of the network are
the number of tiers, the node density, the communication radio coverage radius, the number and
location of the sinks. All these parameters are examined for all possible combinations in detail.
Identifying the significant role of clustering in network lifetime prolongation, the rest of the paper
is organized as follows. In Section 2, we briefly describe the idea of clustering by examining a
rich number of studies conducted on this topic. In Section 3, we provide information about the
methods and architectures utilized during simulations. Section 4 is devoted for graphical
presentation of the simulation results and discussion. Lastly, in Section 5 we provide concluding
remarks.
2. RELATED WORK
Clustering is virtually slicing the network topology into grids (Figure 1) and grouping the sensor
nodes under these grids according to a number of benchmarks.

3
Figure 1. Voronoi based clustering
One of the nodes in each cluster is charged with being cluster head (CH). Other nodes in the
cluster which are called plain nodes gather data from the environment and deliver it to the CH
node. CH is responsible for conveying the overall data gathered in its cluster to the sink. In
traditional non-cluster based sensor networks, each sensor node gathers data from physical
environment and aims to transmit its data to the sink somehow. If it is thought that all the nodes
in the topology try to deliver their data simultaneously by flooding, a huge amount of data
transmission will occur. Besides, due to the fact that all the nodes try to access the common
transmission media at the same time, serious delays will occur as a result of collision prevention
mechanisms. Moreover, in consequence of routing loops and multiple routes, redundant energy
consumptions will result. Therefore, in terms of preventing redundant energy consumption during
data transmission, clustering approach provides very significant gains by means of simplifying
the communication and enhancing the scalability [11]. The objective is to find the optimum
method of organizing the nodes into clusters and electing the most appropriate node as the CH in
each cluster in order to achieve energy efficiency by realizing load balance among the nodes.
Many studies have been proposed about cluster-based WSNs. LEACH [12-13], is one of the first
and fundamental studies conducted on WSNs and has led to many subsequent studies about
clustering. The lifetime of the network is partitioned into rounds in LEACH. Cluster formation is
done in an autonomous and distributed manner by the nodes without centralized supervision.
Each round is divided into two phases: set-up and data transmission states. At set-up phase, each
node in the topology holds a random number and depending on this number is elected to be a
cluster head. Load is evenly distributed by rotating the charge of being CH among all nodes.
Thus, the drainage of the nodes in the battery is delayed. Another impressive solution proposed in
LEACH is the CHs making data aggregation in order to reduce the amount of data to be
transmitted.
Another study of clustering following LEACH is PEGASIS [14]. Although PEGASIS is
perceived as an improvement of LEACH, its basic principle is based on the chain structure rather
than a cluster scheme.
HEED [15], is another work which achieves considerable improvements on energy conservation
in WSNs. As in LEACH, CH selection is done periodically but not at each round. In contrast with
LEACH, CH selection is not done randomly, but is rather made according to a hybrid parameter
which is a combination of the residual energy levels of the nodes and a cost value called the
average minimum reachability power (AMRP). AMRP is the total energy consumed by all the
other nodes in the cluster if the aforementioned node becomes CH.
Clustered routing for selfish sensors (CROSS) [16] and its improved version localized game
theoretical clustering algorithm (LGCA) [17] is based on the game theory for cluster formation
and CH election. CROSS depends on global knowledge about the topology which is neither

4
practical nor realistic. In contrast, LGCA employs localized information which is more suitable
for energy poor WSNs.
Zhu et al. have proposed an architecture [18] in which clustering is basically performed by
utilizing Hausdroff Distance [19]. The first criterion that is considered during CH selection phase
is the residual energy level of the nodes. Secondly, if the residual energy levels are equal, then the
proximity off the nodes is taken into account. Inter-cluster routing is performed by means of
utilizing classical Bellman-Ford’s shortest path approach [20].
In EECS (An Energy Efficient Clustering Scheme in Wireless Sensor Networks) [21], the
residual energy levels of the nodes are again considered. Another factor impacting the CH
selection is the distance between the CH candidates and the sink because inter-cluster
communication is performed directly between the CHs and the sink.
In order to prevent redundant message exchange suffered during CH election phase, Cui et al.,
proposes an efficient idea which is called passive clustering [22]. With passive clustering, each
CH candidate determines a random waiting time inversely proportional with the residual energy
levels of the nodes. That is, a node with low energy level waits a longer time to announce its
leadership. Therefore, the timers of the nodes with higher energy levels expire earlier and
announce their leadership before the others. Thus, other nodes hearing the announcement give up
the competition.
Inter-cluster communication is another challenge to be considered in cluster-based networks. Data
delivered at the end of the intra-communication phase must be conveyed to the sink by the CH.
This can be achieved either by single-hop or multi-hop communication. ANCAEE (A Novel
Clustering Algorithm for Energy Efficiency in Wireless Sensor Networks) offers single-hop
transmission for intra-cluster communication and multi-hop transmission for inter-cluster
communication [23].
In addition to the studies mentioned above there have been several other studies on cluster-based
sensor networks [24-28]. The next section gives details about the methods and architectures
utilized in the system while analyzing the impacts of different structural variations on energy
consumption.
3. EXPERIMENTAL ARCHITECTURE AND DETAILS
This section outlines the methodology and some significant concepts utilized in our analysis. We
performed a large set of simulations with various combinations of node density, tier count, sink
settlement, radio coverage, and cluster sizing. Each simulation was run until the first node death
which defines the network lifetime. For performance measurement, we considered the network
lifetime, since the primary challenge to be accomplished for WSNs is prolonging the working life
of the network.
Instead of considering an event-based system, our simulations are based on the scenario that all
nodes in the topology periodically gather data and try to transmit that data to the sink(s). For
convenience, sensor nodes are assumed to be fixed and randomly distributed in a two-
dimensional plane. Since all nodes potentially participate during inter-cluster communication,
they do not apply any sleep-wake-up schedule.
Details about the main figures utilized during simulations are described in the following
subsections.

5
3.1. Energy Consumption Model
In this paper, the classical energy model as described in LEACH is used. As is known, primary
factors affecting energy consumption are the number of bits transmitted and the distance between
the communicating pairs. If the distance between the communicating nodes is greater than the
threshold value, then the impact of the distance on the energy consumption grows exponentially
as shown in Eq. (1-3).
Esnd(l,d) = Esnd-elec(l) + Esnd-amp(l,d) (1)
Esnd(l,d)=
(l * Eelec) + (l * ℇfs * d2
) , d<do
(2)
(l * Eelec) + (l * ℇmp * d4
) , d≥do
Ercv = l * Eelec (3)
3.2. Network Lifetime
Several network lifetime definitions are proposed in the literature [29-34]. Some of them consider
the time in which a certain amount of the nodes die. Another idea to consider is the time after
which there is a region no longer covered by the network. The one that makes the most sense and
which we applied in this study is the time when the first node fails. When a node dies, it would
neither be accurate nor realistic to assume that the remaining network will work well. Eventually,
the node is dead and no data can be obtained from the area for which the dead node is
responsible. Besides, this can result in a network partition situation which means there are two
nodes which no longer can communicate with each other.
3.3. Cluster Head Election
Cluster Heads (CHs) have the responsibility of relaying the aggregated data of the corresponding
cluster to the sink. Therefore, this heavy mission should be shared among different nodes as much
as possible. Otherwise, the node assigned as CH drains its battery quickly. In this study, three
types of CH election methods are analysed:
CH Election Model 1: Every node in a cluster runs the same algorithm similar to the one
proposed in [22]. The result of the algorithm is a time value that determines the access time of a
node to the common media for announcing its leadership. Other nodes hearing this announcement
give up the election process and assign that node as the master node. Calculated waiting time
(Tw(i)) at each node is reversely proportional with the distance of the node to the centre of the
corresponding cluster and the residual energy level of that node:
Tw(i) = d(node(i), ClsCentre) / EngRes_node(i) (4)
where:
d(node(i), ClsCentre) is the Euclidean distance between node(i) and the centre point of the cluster
to which it belongs;
EngRes_node(i) is the residual energy of node(i).
According to Equation (4), nodes positioned around the cluster centre with higher residual energy
levels wait shorter durations and therefore have a higher probability of being elected as CH than
the others.
CH Election Model 2: This model uses a similar method to the one identified in model 1. This
time, an extra parameter is involved during CH election phase as presented in Eq. (5). Nodes
deployed between the centre of the corresponding cluster and the sink can be a CH. Furthermore,

6
the distance between the node and the target sink is considered rather than the distance from the
node to the centre of the cluster (Eq. (6)).
isCHCnd = (5)
where:
isCHCnd is a value that determines the possibility of a node to be a CH;
node(i)ypos, and ClsCentreypos denote the (y) coordinates of node(i) and the centre point of the
cluster it belongs to respectively.
Obviously, according to Eq. (6), CHs are elected among the nodes that are positioned between the
centre point of the corresponding cluster and the target sink.
Tw(i) = (isCHCnd*d(node(i),TrgSink))/EngRes_node(i) (6)
where:
isCHCnd is a value that determines the possibility of a node to be a CH;
d(node(i), TrgSink) is the Euclidean distance between node(i) and the target sink;
EngRes_node(i) is the residual energy of node(i).
CH Election Model 3: In this method, every node in the cluster can be elected as a CH. There is
no constraint like the one defined in Model 2. Again, the target is the sink(s) and nodes closer to
the sink(s) with more residual energies have a greater chance to be elected as CH (Eq. (7)).
Tw(i)= d(node(i),TrgSink) / EngRes_node(i) (7)
3.4. Routing
Next-hop selection is performed depending on the geographical positions of the nodes. It is
assumed that all nodes are aware of their relative two dimensional coordinates in the topology.
Furthermore, they are assumed to be informed about the coordinates of their neighbours and the
sinks settled in the topology. A number of techniques have been proposed in the literature about
localization and positioning concepts. The first coming to mind is that equipping the sensor node
with a Global Positioning System (GPS) receiver. However, that is not a promising solution
because of deployment and cost limitations. There are other alternative solutions proposed such
as lateration and angulation techniques [35]. Since it is out of scope of this study, no specific
positioning method is studied in the paper.
3.5. Intra-Cluster Communication
Data gathered by each plain node in the cluster is delivered directly to the CH in a single-hop
manner if it is in the coverage area of the sender node. Otherwise, multi-hop transmission is
utilized. The packet emerging from the plain sensor node is forwarded to one of the neighbours
belonging to the same cluster which is closest to the CH. The next-hop selection method for intra-
cluster packet transmission is given below:
Algorithm 1 Intra-Cluster Next-Hop Selection Method
findNextIntraClsHop(){
if (isInCov(this, CH)) then
sendPckDirectlyToCH()
end if
else
distance ← ∞
for (i←1 to numOfNgbs ) do
if (d(this,CH) < dist(ngb(i),CH)) then
if (isInSameCls(this,ngb(i))) then

7
if (dist(ngb(i),CH) < distance) then
distance = dist(ngb(i),CH)
nexthop = ngb[i]
end if
end if
end if
end for
sendIntraClsPckToNxtHop(nexthop)
end if
}
3.6. Inter-Cluster Communication
Data aggregated at each cluster is delivered by the CHs directly to the closest sink if the sink is in
the coverage area. Otherwise, the aggregated packet is forwarded to the node that is closest to the
target sink. If noticed, the next hop candidate of the inter-cluster packet is not required to be in
the same cluster. Next-hop calculation method for Inter-cluster communication is given below:
Algorithm 2 Inter-Cluster Next-Hop Selection Method
findNextInterClsHop(){
TrgSink ← calcTrgSink();
if (isInCov(this, TrgSink)) then
sendPckDirectlyToTrgSink(TrgSink)
end if
else
distance ← ∞
for (i←1 to numOfNgbs ) do
if (d(this,TrgSink) > dist(ngb(i),TrgSink)) then
if (dist(ngb(i),TrgSink) < distance) then
distance = dist(ngb(i), TrgSink)
nexthop = ngb[i]
end if
end if
end for
sendInterClsPckToNxtHop(nexthop)
end if
}
3.7. Packet Structures
Length of an Intra-cluster packet is 52 bytes that takes 1625µs to transmit with the utilized radio
data rate. As is known, WSNs are data-centric applications, not id-based like other traditional
networks. That is, data collection centre does not deal with the ID of the data source. It is only
concerned with the content. ID is only needed during forwarding operations inside the topology.
Thus, there is no need to apply conventional, redundant IP or MAC addresses during in-network
forwarding. It is sufficient to define short in-network unique addresses for forwarding purposes.
Since all nodes are assumed to be aware of their geographical positions of themselves and their
neighbours, and it is also assumed that two distinct nodes do not overlap, these relative two-
dimensional coordinates constitute the ID of the nodes. Intra-cluster packet structure is presented
in Figure 2.
Figure 2. Intra-Cluster packet structure

8
CHs aggregate the data delivered by the plain nodes in its corresponding cluster and generate an
Inter-Cluster packet to be transmitted to the sink. Structure of an Inter-Cluster Packet is depicted
in Figure 3.
Figure 3. Inter-Cluster packet structure
As shown in Figure 3, the first 6 octets of the Inter-Cluster packet are fixed for every Inter-
Cluster packet. ClsID and CHID slots contain the IDs of the owner cluster and the corresponding
CH. In this way, when the aggregated packet arrives at the data collection centre, this information
identifies the region to which that packet belongs. For each round, a new sensor node is elected as
a CH. Therefore, the value of the second octet part changes in each round. Field TrgSinkID
denotes the ID of the target sink. Like the sensor nodes, relative two-dimensional coordinates of
the sinks constitutes their IDs. Topology is virtually divided into clusters uniformly and
permanently and each cluster defines its target sink at the beginning of its lifecycle, that is, the
closest one relative to the centre point of the cluster. Remaining parts of the Inter-cluster packet
comprises of the data delivered by each plane node in the cluster. Since the number of nodes
varies for each cluster, a general formula identifying the total length of an Inter-Cluster packet is
as follows:
LngthInterClsPck = 48 +(LngthIntraClsPck * NumOfNodes) (8)
3.8. Cluster Size
One of the most important challenges encountered in WSNs is the hot-spot problem. As stated
above, since sensor nodes are very tiny devices, their resources have limited capacities. The same
limitation is also valid for the communication coverage radius. Sensor nodes far from the sink
cannot transmit their data directly to the sink. Moreover, conveying the data directly which is
actually single-hop transmission is not preferred because the energy consumed during data
transmission is exponentially proportional by the distance. Therefore, multi-hop communication
is preferred in WSNs. Though multi-hop communication seems to be advantageous, another vital
challenge to be considered is called the hot-spot problem. Nodes closer to the sink act like a relay
and convey the data incoming from the remote nodes to the sink as shown in Figure 4. Hence, all
of the data traffic passes over a limited number of nodes that will cause these nodes to quickly
drain the battery, which is called the hot-spot problem.
Figure 4. Hot-spot problem
One of the promising proposals for solving the hot-spot problem is the unequal clustering method.
Researchers claim that forming the cluster located closer to the sink with smaller sizes and the
remote ones with larger sizes provides considerable gains in terms of energy conservation [36-
42].
In this part, we inspect the effect of cluster sizes on energy consumption for three types of
methods:

9
- Clusters closer to the sink with smaller sizes and remote clusters with larger sizes
(ClsSizeModel1)
- Clusters closer to the sink with larger sizes and remote clusters with smaller sizes
(ClsSizeModel2)
- All clusters with equal size (ClsSizeModel3)
3.9. The Number and Location of the Sinks
Generally, researchers locate a single sink at one side or at the centre of the topology. However,
position and number of the sinks can affect the energy consumed in the topology. Therefore,
another important point examined in this paper is the variation in network lifetime depending on
the position and the number of the sink(s) in the network. Simulations were performed according
to three different sink(s) localization types.
- Sink(s) located at one side of the topology (SinkPositionModel1)
- Sink located at the centre of the topology (SinkPositionModel2)
- Sink(s) positioned around the topology (SinkPositionModel3)
4. EXPERIMENTAL RESULTS
In this section, we analyse the performance and the impact of five structural factors on energy
consumption: the number of tiers, the cluster sizes, the number and the location of the sinks, the
node density and the radio coverage. While analysing the effects of these factors, different types
of CH election methods are utilized and performances are compared. The simulations are
performed on a 500*500 squarely shaped area where the nodes are randomly deployed. Lifetime
is comprised of periodic rounds that each consists of CH election, data aggregation, intra-cluster
data transmission and inter-cluster data communication phases. Topology is virtually divided into
tiers and clusters.
In order to prevent common media access collisions, a MAC protocol similar to 802.11 with
RTS/CTS mechanism is used. As mentioned in the previous section, we employed the classical
energy calculation model that depends primarily on the distance. Parameters utilized during
simulations are given in Table1.
Table 1. Simulation parameters
Radio transmission data rate 250 Kbps
d0 (threshold distance) 85 m
R0 (coverage radius) 100 m
Eelec 50 nJ/bit
εfs 10 pJ/bit/m2
Emp 0.0013
pJ/bit/m4
As mentioned earlier, the impacts of cluster sizing, the number and the location of the sink(s), the
number of tiers and the node density on energy consumption are represented comprehensively.
Simulations are performed under three structural titles depending on how the clusters are sized.
As clarified previously, these are: clusters closer to the sink with smaller sizes and remote clusters
with larger sizes (ClsSizeModel1), clusters closer to the sink with larger sizes and remote clusters
with smaller sizes (ClsSizeModel2) and all clusters with equal sizes (ClsSizeModel3).
4.1. Clusters Close to the Sink with Smaller Sizes (ClsSizeModel1)
Another alternative solution proposed by the researchers is to slice the topology into tiers. As
shown in Figure 5, incrementing the number of tiers redundantly makes a negative impact on the

10
lifetime of the network. With a topology of containing 500 nodes, the best performance is
provided by utilizing 2 tiers. SelectCH_Centre, SelectCH_EN_AfterCentre and
SelectCH_EN_withNoCons correspond to the CH election schemes CH Election Model 1, CH
Election Model 2 and CH Election Model 3 respectively.
Figure 5. Change in the network lifetime depending on the number of tiers
Another possible factor that may affect the network lifetime is the number of nodes in the
network that is node density. By holding the size of the topology area constant, increasing the
number of nodes increases the node density. However, more nodes means more data packets to be
transmitted. Thus, the load is observed to be shared by the time the density increases; however,
increase in the network traffic balances this factor. Figures 6-8 show the lifetime performance
depending on the node density for the networks that the sink(s) positioned one side, sink at the
centre and sink(s) surrounding the nodes respectively.
Figure 6. Change in the network lifetime depending on the node density for (SinkPositionModel1,
ClsSizeModel1) pair
ClsSizeModel1) pair

11
ClsSizeModel1) pair
Figures 6-8 clarify that a topology model surrounded with the sinks provides the best
performance in terms of network lifetime.
Another possible factor that can affect the network performance is the number of sinks in the
system. We examined whether an increase in the number of sinks prolongs the lifetime of the
network.
Figure 9. Change in the network lifetime depending on the number of sinks for (SinkPositionModel1,
ClsSizeModel1) pair
ClsSizeModel1) pair
Figures 9-10 reprove that with SinkPositionModel3, network lifetime is doubled by increasing the
number of sinks in the network. Since, the sensor nodes have limited coverage capacities and the
energy consumption is proportional with the communication distance, multi-hop communication
is preferred in WSNs. In our simulation model, CHs directly forward the aggregated data to the
sink(s) if they are in the communication range. However, this causes an extra burden. As

12
mentioned in the previous section, energy consumption of the communication unit is
exponentially proportional with the distance between the receiver and the sender. This idea is
supported by Figures 11-13. Radio coverage improvement does not bring an extra advantage in
terms of energy conservation.
Figure 11. Change in the network lifetime depending on the radio coverage for (SinkPositionModel1,
ClsSizeModel1) pair
ClsSizeModel1) pair
ClsSizeModel1) pair
4.2. Clusters Close to the Sink with Larger Sizes (ClsSizeModel2)
Figures 14-21 present the changes occur in the network lifetime depending on the parameters
presented in the previous section. This time, clusters closer to the sink(s) have larger sizes and the
further ones with smaller sizes.

13
ClsSizeModel2) pair
ClsSizeModel2) pair
ClsSizeModel2) pair

14
ClsSizeModel2) pair
ClsSizeModel2) pair
ClsSizeModel2) pair
ClsSizeModel2) pair

15
ClsSizeModel2) pair
4.3. All Clusters with Equal Sizes (ClsSizeModel3)
Finally, Figures 22-29 present the changes that occur in the network lifetime depending on the
parameters presented in the previous section. This time, all the clusters in the network have equal
sizes.
ClsSizeModel3) pair
ClsSizeModel3) pair

16
ClsSizeModel3) pair
ClsSizeModel3) pair
ClsSizeModel3) pair
ClsSizeModel3) pair

17
ClsSizeModel3) pair
ClsSizeModel3) pair
5. CONCLUSIONS
The primary energy-consuming unit of a sensor node is the communication unit. It is crucial that
while designing protocols, methods, and architectures for WSNs, this energy constraint problem
should be considered. Up to now, several research activities performed and various methods have
been proposed about cluster-based WSNs. This paper presents a brief analysis of the effects of the
various structural factors in terms of energy consumption in cluster-based WSNs. General belief
about cluster-based WSNs is that in order to alleviate the hot-spot problem, clusters located near
the sink should be smaller-sized than the remote ones. Other possible factors that may affect the
lifetime of the network are the number of tiers, the node density, the communication radio
coverage radius, the number and location of the sinks. All these parameters are examined for all
possible combinations in detail. Depending on our simulations, the best performance in terms of
the network lifetime is provided by positioning the sinks around the network. Increasing the node
density up to a level is another factor that affects the energy consumption positively. Also, it is
proved that sizing the clusters closer to the sink smaller than those further away enhances the
energy conservation. Furthermore, it is also clarified that larger radio coverage does not have a
definite positive effect in terms of energy conservation.
REFERENCES
[1] Al-Sabbagh A.A.J., Alsabah R.A.S., Mjhool, A.Y., “Effects of Mobility Models and Nodes
Distribution on Wireless Sensors Networks”, International Journal of Ad hoc, sensor & Ubiquitous
Computing, vol. 5, no. 6, pp. 1-8, 2014. doi:http://dx.doi.org/10.5121/ijasuc.2014.5601
[2] Chen Z., Xiao Y., Li X., Li R., “Shortest Path Algorithm for Data Transmission in Wireless Ad-hoc
Sesnsor Networks”, International Journal of Ad hoc, sensor & Ubiquitous Computing, vol. 5, no. 4,
pp. 13-21, 2014. doi:http://dx.doi.org/10.1155/2013/829861.
[3] Cevik T, Zaim A. H., Yıltaş D., “Localized power aware routing with an energy efficient pipelined
wakeup schedule for wireless sensor networks”, Turkish Journal of Electrical Engineering &
Computer Sciences, vol. 20, pp. 964–978, 2012.

18
[4] Cevik T., Zaim A. H., “EETBR: energy efficient token based routing for wireless sensor networks”,
Turkish Journal of Electrical Engineering & Computer Sciences, vol. 21, pp. 513-526, 2013.
[5] Akyildiz F., Su W., Sankarasubramaniam Y., Cayirci E., “A survey on sensor networks”, IEEE
Communications Magazine, vol. 40, no. 8, pp. 102–114, 2002.
[6] Wang H.L., Tsai R.G., “Application of Periodical Shuffle in Controlling Quality of Service in
Wireless Sensor Networks”, International Journal of Ad hoc, sensor & Ubiquitous Computing, vol. 4,
no. 2, pp. 17–30, 2013. doi:http://dx.doi.org/10.5121/ijasuc.2013.4202
[7] Çevik T, Zaim A. H., “A Multichannel Cross-Layer Architecture for Multimedia Sensor Networks”,
International Journal of Distributed Sensor Networks, 2013.
doi:http://dx.doi.org/10.1155/2013/457045.
[8] Cjoe H. J., Ghosh P., Das S. K., “Qos aware data reporting control in cluster based wireless sensor
networks”, Computer Communications, vol. 33, pp. 1244-1254, 2010.
[9] Raghunathan V., Schurgers C., Park S. I., Srivastava M. B., “Energy-aware wireless microsensor
neetworks”, IEEE Signal Processing Magazine, vol. 19, pp. 40-50, 2002.
[10] Kumar V., Jain S., Tiwari S., “Energy Efficient Clustering Algorithms in Wireless Sensor Networks:
A Survey”, International Journal of Computer Science Issues, vol. 8, no. 2, pp. 259-268, 2011.
[11] Abbasi A. A., Younis M., “A Survey on Clustering Algorithms for Wireless Sensor Networks”,
Computer Communications, vol. 30, no.14-15, pp. 2826-2841, 2007.
[12] Heinzelman W. R., Chandrakasan A., Balakrishnan H., “Energy-efficient communication protocol for
wireless microsensor networks”, In Proc. of the 33rd Hawaii International Conference on System
Sciences, 2000.
[13] Heinzelman W., Chandrakasan A., Balakrishnan H., “An application specific protocol architecture for
wireless microsensor networks”, IEEE Transactions on Wireless Communications, vol. 1, pp. 660–
670, 2002.
[14] Lindsey S., Raghavendra C. S., “PEGASIS: power-efficient gathering in sensor information systems”,
In Proc. of the IEEE Aerospace Conference, pp. 1125–1130, 2002.
[15] Younis O., Fahmy S., “HEED: a hybrid, energy-efficient, distributed clustering approach for ad hoc
sensor networks”, IEEE Transactions on Mobile Computing, vol. 3, pp. 366–379, 2004.
[16] Koltsidas G., Pavlidou F. N., “A game theoretical approach to clustering of ad-hoc and sensor
networks”, Telecommunication Systems, vol. 47, no. 1-2, pp. 81–93, 2011.
[17] Xie D., Sun Q., Zhou Q., Qiu Y., Yuan X., “An Efficient Clustering Protocol for Wireless Sensor
Networks Based on Localized Game Theoretical Approach”, International Journal of Distributed
Sensor Networks, 2013. doi: http://dx.doi.org/10.1155/2013/476313.
[18] Zhu X., Shen L., Yum T. P., “Hausdorff Clustering and Minimum Energy Routing for Wireless
Sensor Networks”, IEEE Transactions on Vehicular Technology, vol. 58, no. 2, pp. 990-997, 2009.
[19] Huttenlocher D. P., Klanderman G. A., Rucklidge W. J., “Comparing Images Using the Hausdorff
Distance”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 15, no. 9, pp.850-
863, 1993.
[20] Bertsekas D., Gallager R., Data Networks, 2nd ed. Englewood Cliffs, NJ:Prentice Hall, 1992.
[21] Ye M., Li C., Chen G., Wu J., “EECS: An Energy Efficient Clustering Scheme in Wireless Sensor
Networks”, In Proc. of the 24th IEEE International Performance, Computing and Communications
Conference, pp. 535-540, 2005.
[22] Cui S., Ferens K., “Energy Efficient Clustering Algorithms for Wireless Sensor Networks”, In Proc.
of the International Conference on Wireless Networks (ICWN’11), Las Vegas, NV, 2011.
[23] Abidoye A. P., Azeez N. A., Adesina A. O., Agbele K. K., “ANCAEE: A Novel Clustering Algorithm
for energy Efficiency in Wireless Sensor Networks”, Wireless Sensor Network, pp. 307-312, 2011.
[24] Xu Y., Heidemann J., Estrin D., “Geography-informed Energy Conservation for Ad Hoc Routing”, In
Proc. of the 7th annual international conference on Mobile computing and networking, pp. 70-84,
2001.
[25] Leng S., Zhang Y., Chen H., Zhang L., Liu K., “A Novel k-Hop Compound Metric Based Clustering
Scheme for Ad Hoc Wireless Networks”, IEEE Transactions on Wireless Communications, vol. 8, no.
1, pp. 367-375, 2009.
[26] Azim A., Islam M. M., “Hybrid LEACH: A Relay Node Based Low Energy Adaptive Clustering
Hierarchy for Wireless Sensor Networks”, In Proc. of the IEEE 9th Malaysia International
Conference on Communications, pp. 911-916, 2009.
[27] Youssef M., Youssef A., Younis M. F., “Overlapping Multihop Clustering for Wireless Sensor
Networks”, IEEE Transactions on Parallel and Distributed Systems, vol. 20, no. 12, pp. 1844-1856,
2009.

19
[28] Kumar D., Aseri T. C., Patel R. B., “EECH: Energy Efficient Heterogeneous Clustered Scheme for
Wireless Sensor Networks”, Computer Communications, vol. 32, 662-667, 2009.
[29] Chen Y., Zhao Q., “On the Lifetime of Wireless Sensor Networks”, IEEE Communications Letters,
vol.9, no. 11, pp. 976-978, 2005.
[30] Kang I., Poovendran R., “Maximizing Network Lifetime of Broadcasting over Wireless Stationary Ad
Hoc Networks”, Mobile Networks and Applications (MONET), Special Issue on Energy Constraints
and Lifetime Performance in Wireless Sensor Networks, vol. 10, no. 6, pp. 879-896, 2005.
[31] Shi Y., Hou Y. T., Efrat A., “Algorithm Design for Base Station Placement Problems in Sensor
Networks”, In Proc. of the 3rd Int’l Conf. on Quality of Service in Heterogeneous Wired/Wireless
Networks (QShine’06), 2006.
[32] Vass D., et al., “Energy Efficiency in Wireless Sensor Networks Using Mobile Base Station”, In Proc.
of 11th Open European Summer School and IFIP WG6.6, WG6.4, WG6.9 Workshop (EUNICE
2005), 2005.
[33] Raicu I., et al., “Local Load Balancing for Globally Efficient Routing in Wireless Sensor Networks”,
International Journal of Distributed Sensor Networks, vol.1, no. 2, pp. 163-185, 2005.
[34] Cheng Z., Perillo M., Heinzelman W. B., “General Network Lifetime and Cost Models for Evaluating
Sensor Network Deployment Strategies”, IEEE Trans on Mobile Computing, vol 7, no. 4, pp. 484-
497, 2008.
[35] Karl H., Willing A., Protocols and Architectures for Wireless Sensor Networks, Wiley, 2007.
[36] Yu J., Qi Y., Wang G., Guo Q., Gu X., “An Energy-Aware Distributed Unequal Clustering Protocol
for Wireless Sensor Networks”, International Journal of Distributed Sensor Networks, 2011.
doi:http://dx.doi.org/10.1155/2011/202145.
[37] Lee S., Lee J., Sin H., Yoo S., Lee S., Lee J., Lee Y., Kim S., “An Energy-Efficient Distributed
Unequal Clustering Protocol for Wireless Sensor Networks”, In Proc. of World Academy of Science:
Engineering & Technology, pp. 443-447, 2008.
[38] Liao Y., Qi H., Li W., “Load-Balanced Clustering Algorithm with Distributed Self-Organization for
Wireless Sensor Networks”, IEEE Sensors Journal, vol. 13, no. 5, pp. 1498-1506, 2013.
[39] Chang-Jiang J., Wei-ren S., Min X., Xian-lun T., “Energy-Balanced Unequal Clustering Protocol for
Wireless Sensor Networks”, The Journal of China Universities of Posts and Telecommunications, vol.
17, no. 4, pp. 94-99, 2010.
[40] Bai F., Mou H., Sun J., “Power-Efficient Zoning Clustering Algorithm for Wireless Sensor
Networks”, In Proc. of the International Conference on Information Engineering and Computer
Science, pp. 1-4, 2009.
[41] Li C., Ye M., Chen G., Wu J., “An Energy-Efficient Unequal Clustering Mechanism for Wireless
Sensor Networks”. In Proc. of the IEEE International Conference on Mobile Ad-hoc and Sensor
Systems, 2005.
[42] Soro S., Heinzelman W. B., “Prolonging the Lifetime of Wireless Sensor Networks via Unequal
Clustering”, In Proc. of the 19th IEEE International Parallel and Distributed Processing Symposium
(IPDPS’05), 2005.
AUTHORS
Taner Cevik received the B. S., M.S. and Ph.D. degrees in computer engineering from
Istanbul Technical University in 2001, Fatih University in 2008, and Istanbul
University in 2012 respectively. In 2006, he joined the Department of Computer
Engineering, Fatih University, as a research assistant, and in 2010 became an
instructor at the same university. Since 2013, he has served as an assistant professor at
Fatih University.
Fatih Ozyurt received the B.S. and M.S. degrees in computer engineering from Eastern
Mediterranean University Cyprus, and Fatih University Istanbul Institute of Science in
2011 and 2014, respectively. He continues his education as a PHD student in Software
Engineering at Firat University in Elazig.

IMPACTS OF STRUCTURAL FACTORS ON ENERGY CONSUMPTION IN CLUSTER-BASED WIRELESS SENSOR NETWORKS: A COMPREHENSIVE ANALYSIS

More Related Content

What's hot (14)

Similar to IMPACTS OF STRUCTURAL FACTORS ON ENERGY CONSUMPTION IN CLUSTER-BASED WIRELESS SENSOR NETWORKS: A COMPREHENSIVE ANALYSIS (20)

Recently uploaded (20)

IMPACTS OF STRUCTURAL FACTORS ON ENERGY CONSUMPTION IN CLUSTER-BASED WIRELESS SENSOR NETWORKS: A COMPREHENSIVE ANALYSIS