Hindawi

Journal of Sensors
Volume 2020, Article ID 9040395, 19 pages
https://doi.org/10.1155/2020/9040395

Research Article
Optimized Cluster-Based Dynamic Energy-Aware Routing
Protocol for Wireless Sensor Networks in Agriculture Precision

Kashif Naseer Qureshi ,1 Muhammad Umair Bashir,1 Jaime Lloret ,2 and Antonio Leon2
1
Department of Computer Science, Bahria University, Islamabad, Pakistan
2
Universitat Politecnica de Valencia, Spain

Correspondence should be addressed to Kashif Naseer Qureshi; [email protected] and Jaime Lloret; [email protected]

Received 28 October 2019; Accepted 2 December 2019; Published 31 January 2020

Guest Editor: Zhenxing Zhang

Copyright © 2020 Kashif Naseer Qureshi et al. This 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.

Wireless sensor networks (WSNs) are becoming one of the demanding platforms, where sensor nodes are sensing and monitoring
the physical or environmental conditions and transmit the data to the base station via multihop routing. Agriculture sector also
adopted these networks to promote innovations for environmental friendly farming methods, lower the management cost, and
achieve scientific cultivation. Due to limited capabilities, the sensor nodes have suffered with energy issues and complex routing
processes and lead to data transmission failure and delay in the sensor-based agriculture fields. Due to these limitations, the
sensor nodes near the base station are always relaying on it and cause extra burden on base station or going into useless state.
To address these issues, this study proposes a Gateway Clustering Energy-Efficient Centroid- (GCEEC-) based routing protocol
where cluster head is selected from the centroid position and gateway nodes are selected from each cluster. Gateway node
reduces the data load from cluster head nodes and forwards the data towards the base station. Simulation has performed to
evaluate the proposed protocol with state-of-the-art protocols. The experimental results indicated the better performance of
proposed protocol and provide more feasible WSN-based monitoring for temperature, humidity, and illumination in
agriculture sector.

1. Introduction watch to track and monitor the dangerous situation and

remain alert against terrorist attacks [5]. In agriculture, sen-
Precision agriculture refers to a science using advance tech- sor nodes are deployed to sense the temperature, pressure,
nologies to provide cost management, crop growth, and pro- humidity, and wind speed. In addition, the sensor nodes also
duction in agriculture fields. One of the major driver of sense environmental conditions for weather forecast and
agriculture precision is wireless sensor networks (WSNs) natural disaster happening probability. In these networks,
where the sensor nodes are monitoring the physical or envi- the sensor nodes are categorized into coordinator and nor-
ronmental conditions including humidity, temperature, and mal nodes to collect the data from the agricultural field [6].
illumination and send the sensed data to the base station The sensor nodes sense the required parameters and analyze
(BS) via single-hop or multihop coordinator nodes [1–3]. the distance threshold (dTh ) and then forward the sensed
This technology has various beneficial applications in other data to the sink node by single-hop or multihop communica-
fields like healthcare, military, transportation, security, and tion. The role of the sink node is to collect the data from sen-
agriculture. In healthcare, sensor nodes have deployed to col- sor nodes and further transmit to gateway or BS and then
lect the patient physiological or biometric information such further send to the central management system for decision
as ECG, heart rate, and blood pressure [4]. In the military, making as shown in Figure 1.
sensor nodes are deployed to track the soldiers on the battle- Sensor nodes are small in size with low computational
field, for monitoring, find the location of platoons, and pro- power and energy resources [7]. Sensor nodes are used for
tect the forces. In security, sensor nodes can offer a careful monitoring the environmental conditions like crop conditions
2 Journal of Sensors

Internet

Sink

Gateway

User Sensor nodes

Wireless
sensor node

Figure 1: Architecture field with WSN deployment.

and other environmental parameters. The sensor nodes are protocol flat and hierarchical. In flat routing protocols, all the
deployed on the surface of soil or inside the soil. There are nodes in the network play an identical role. The main issue in
different technologies and standards which have been flat routing protocols is scalability, load balancing, route
adopted based on applications and data rate, frequency band, maintenance, and not feasible for the large networks
power consumption, and distance. Some common technolo- [20, 21]. To address the scalability and load balancing issues
gies are Wibree, Wifi, GPRS, WiMAX, Bluetooth, and ZigBee in flat routing, hierarchical routing protocols are introduced.
[2, 8]. Monitored data was obtained from the deployed sen- It is also called cluster-based routing, in which all sensor
sor nodes and then was wirelessly forwarded to the BS for nodes in the network are separated into layers based on resid-
data collection. The BS initiates the decision for further pro- ual energy and assigned the different roles. In the entire net-
cesses. Users received the crop growth information or other work, the sensor nodes are divided into a group called
information related to the drip irrigation and take further ini- clusters [22]. Each cluster has cluster members (CMs) and
tiatives to improve the microenvironment for their product one cluster head (CH). The CH is responsible for coordina-
[9]. In agriculture, for achieving the precision control, the tion within the cluster and forwarding the data to other
sensor nodes monitored different parameters, analysis of CHs or BS. Hierarchical routing protocols or clustering
monitored data for decision making and applying the control protocols are helpful especially for large-scale agriculture
mechanism [10, 11]. There are various efforts to improve the precision-based WSN. It utilized fewer resources, save more
cultivation in agriculture, precision farming, collecting, and energy of sensor nodes, scalable, less packet overhead, and
sending the monitored data [12, 13]. The monitored data is efficiently balances the load among the network as compared
about environmental conditions including weather, wind to flat routing protocol [23–25].
speed, temperature, soil humidity, chemical and physical Complex routing processes and data transmission are the
properties of soil like the pH level, crop identification, leaf main causes of energy depletion among sensor nodes in agri-
area index, leaf moisture content, and weed-disease detec- cultural precision WSN [26, 27]. Aiming at a higher energy
tion. There is another way in which the sensor nodes efficiency for the entire network, a new protocol named
captured the images of fruits, for automated harvesting, Gateway Clustering Energy-Efficient Centroid (GCEEC)
and predicted the soil moisture and organic contents routing protocol is proposed to manage the energy resources.
[14, 15]. Mobility-based sensor nodes are used to mea- The main contributions of this paper is to minimize the
sure the plant mass of crops and analyze the fertilization energy consumption of sensor nodes and to reduce the load
characteristics for best production. Soil strength measure- on CHs. The proposed protocol selects and rotates the CH
ment and prediction-based harvesting time are evaluated on efficient location, i.e., near the energy centroid position
through special sensors [16, 17]. in the cluster to reduce the energy consumption of sensor
In addition, most of the agriculture precision WSN-based nodes in cluster and maximize the CH coverage. Further-
applications need in time and reliable data communication in more, the protocol selects gateway node in cluster to facilitate
the network. Due to limited battery resources, sensor nodes the CH in agriculture environment and significantly reduces
are not able to maintain their operations; recharging and the load on CH.
replacement of batteries are not possible especially in dense The main objectives of this paper are as follow:
forests and large areas [18, 19]. For data communication,
the routing protocols are used to maintain the load balancing (i) To minimize the energy consumption and load
and maximize network lifetime. There are two main types of balancing of the CH by the help of gateway node
Journal of Sensors 3

(ii) Edge node becomes a gateway node to receive more approach having both characteristics of proactive (PEGASIS)
than one joining message from adjacent CHs and reactive (DSR) approach. Utilization of directional trans-
mission scheme helps reduce the communication distance
The rest of the paper is organized as follows: Section 2 which ensures energy efficiency. Furthermore, a trust list is
presents the related work in the area of agriculture-based generated by each node to avoid acknowledgment of receiv-
WSN and its existing energy-based routing protocols. Section ing packets; this will be updated at each round and randomly
3 presents the proposed work design and all steps including checked at any time. Besides this, PDROP also adopts a
flow chart and algorithm. Section 4 presents the experimental Genetic Algorithm (GA) and Bacterial Foraging Optimiza-
results and analysis with state-of-the-art protocols. Last tion (BFO) to discover the optimized path. However, com-
section concludes the paper with future direction. plex routing processes consume more energy and have a
serious impact on the network.
2. Related Work
2.2. Hierarchical Routing Protocol. It is also called clustering
This section discussed the existing energy-efficient routing routing protocols. In these protocols, the whole network
protocols for agriculture precession-based WSN and nodes are divided into a group of nodes called clusters. Each
critically analyzed to find their limitations. Energy-efficient cluster selects CH node which is responsible for transmitting
routing protocols are categorized into two categories, flat the data to the BS.
routing protocol and hierarchical routing protocols which In [32], the authors proposed a Mobile Sink-based Adap-
are discussed in detail. tive Immune Energy-Efficient clustering Protocol (MSIEEP)
and addressed the energy hole problem. The protocol uses
2.1. Flat Routing Protocol. In flat routing, all nodes in the net- Adaptive Immune Algorithm (AIA) to find the sojourn path
work have the same role and perform the same tasks [24]. for the mobile sink. Moreover, the algorithm also finds the
In [28], the authors proposed the dynamic distributed optimize number of CHs based on their dissipated energy
framework protocol known as Energy and Trust-Aware and favorable location. AIA acts as a guide of the mobile sink.
Mobile Agent Migration (ETMAM), in which a mobile agent The significance of mobile sink is to collect the data from the
is used to making route among sensor nodes for data aggre- isolated region of the CH which improved the connectivity of
gation based on energy and trust metric evaluation. A mobile the network. The protocol does not fully address the hole
agent is a self-determined software agent that can move problem due to load balancing issue. In [33], the authors pro-
autonomously among sensor nodes and carry the data for posed a distributed clustering algorithm, namely, Delay-
aggregation. To protect a mobile agent from malicious sensor Constrained Energy Multihop (DCEM) in which CH is
nodes, ETMAM framework provides trust evaluation to the selected in a distributed manner. BS initiates the protocol
mobile agent and bypass the malicious sensor nodes. Fur- by broadcasting ADV message among network sensor nodes;
thermore, the framework also provides optimize migration therefore, each node calculates the distance between itself
route based on energy metrics as well as cloning method to and BS using receive signal strength technique. After that,
aggregate the data from the sensor node. However, the pro- every sensor node broadcasts the advertisement message that
posed framework supports small route mobile agent and is contains its ID and energy level to its neighbor sensor nodes
where response time is low. Power-Aware Heterogeneous so that every neighbor node on receiving advertisement mes-
AODV (PHAODV) in [29] was proposed for the resource sage compares its energy level with energy level information
that should be utilized efficiently. In this protocol, the in receiving advertisement message. If the energy level is
optimized routing path is created by considering the energy greater, then the sensor node becomes candidate CH; other-
status of every sensor node to achieve the load balancing wise, it remains a cluster member. Similarly, the candidate
among heterogeneous networks. The path which consumes CH elects by broadcasting an advertisement message proce-
the least energy is selected as a routing path for data com- dure and becomes CH. The candidate CH with the same
munication from the existing path in the routing table. energy level is further proceeded by computing the trade-
Therefore, all the sensor nodes are keeping aware of the off energy and delay (TED) value. After computing, the
instantaneous change in energy level. Furthermore, link- candidate CH waits for the TED value to receive an advertise-
aware dynamic threshold prevents from route exhausting ment message otherwise becomes the CH. Furthermore, the
and reduces the route error message. However, this proto- DCEM protocol uses intercluster multihop routing cost func-
col has more overhead which leads to energy depletion tion to achieve a minimum cost route from CH to BS. DCEM
issues in the network. does not consider the optimal location of the CH in cluster
An Optimal Base Transmission Strategy (OTDS) [30] is intercluster multihop routing among CH which consumes
proposed in which transmission distance is calculated to bal- more energy.
ance the energy consumption of the entire network. Data In [23], the authors proposed the PSO-ECHS (Particle
mule concept is proposed in which data is collected from sen- Swarm Optimization-Energy Efficient-based Cluster Head
sor nodes and transmits to the BS. Data mule is a mobile Selection) protocol that enhanced the network lifetime. In
node having sufficient storage and energy and collects the the PSO-ECHS algorithm, the CH is selected by fitness func-
data from sensor nodes while roaming across the sensor field tions that consider the distance between sensor node and BS,
and sends it to the BS. PEGASIS-DSR Optimized Routing as well as sensor node and neighbor nodes, and the residual
Protocol (PDORP) is proposed in [31] based on a hybrid energy of sensor nodes. By a minimum value of fitness
4 Journal of Sensors

function, the CH selected and start cluster formation by based on residual energies of the nodes. In addition, CAMP
broadcasting the joining message. Each sensor node after also adjusts the tuning factors including remaining energy,
receiving and joining messages calculates the joining weight node degree, and distance towards the sink node. However,
value. The sensor node joins the CH which has the highest with many benefits, this protocol has significant delay due
joining weight value. In [34], the authors proposed the to its energy calculation and randomly selection of CH in
Energy-Efficient Centroid-based Routing Protocol (EECRP) the network.
for data routing using wireless sensor devices. The term “cen- All the discussed studies mainly focused on energy-
troid” is the mechanical engineering term which means the efficient routing for WSN that reveal the strength and limita-
imaginary central point of mass concentration. Initially in tions that lead to the development of the research problem.
protocol, the BS computes the energy centroid position Based on the literature review, it is revealed that the CH has
among the network and divides the network into a cluster a heavy responsibility for data transmission of the cluster
based on energy centroid position. The node near the energy data towards the BS directly or relaying through other CH.
centroid position is selected as the CH. At the time of CH The CH which directly sends data towards the BS consumes
rotation, the CH recomputed the energy centroid position more energy. The CH far from the BS required more energy
and the node which is near to the energy centroid position in transmitting cluster data towards the BS in a single hop.
elected as the next CH. Furthermore, the protocol also fixed Consequently, these issues lead to the early energy depletion
the threshold distance called MAX distance between the of CH’s which are far from the BS. Moreover, in many
CH and the BS where the CH transmits the data to the CM. schemes such as DUSOC [35], and DCEM [33], CAMP
If CH and BS distance are less than MAX distance, then the [37] CH sends the data towards the BS via intercluster multi-
CH stores the information in the cache and deliver to the hoping. The CH near the sink continuously forwards the CH
next elected CH at the time of CH rotation. data towards the BS. Therefore, uneven load distribution
In [35], the authors proposed the Distributed Unequal among CHs tend to deplete their energy resources rapidly
Size Optimize Cluster (DUSOC) base technique to resolve which leads to disrupt the data dissemination process and
the load balancing issue in the CH. According to the proto- generate routing holes. The CH node selection and CH
col, the BS elects the CH node based on an energy level as well responsibility rotation are one of the most important fea-
as the distance from BS. The CH near the BS chooses the least tures. Therefore, network coverage of CH among cluster
number of sensor nodes as compared to the CH which is far nodes reduces and consumes more energy for data transmis-
away from the BS during the cluster formation stage. Fur- sion to their CH. The optimal location of CH is an important
thermore, intercluster multihop routing among the CH factor which enhances the network coverage among clusters.
approach is adopted for data transmission towards the BS. The optimal location of CH must consider the position where
In [36], the authors proposed the Mobile Energy-Aware energy density nodes found so that the CH responsibility
Cluster-Based Multihop (MEACBM) routing protocol in rotation is must among the nodes that are rich in energy. It
which heterogeneous WSN is divided into clusters, selecting is discussed above that most of the existing clustering
the CH with the highest residual energy. Furthermore, the schemes such as DCEM [33] must improve their intercluster
protocol maintains the coverage and connectivity in the net- multihoping process to overcome load on the CH. Table 1
work by constructing a subcluster for nodes that deployed far presents the protocol comparison in terms of their strategies
away in the network and compute the multihop route for and limitations.
interclustering combination among clusters and subclusters.
After selecting CH, the algorithm divides the network into 3. Gateway Clustering Energy-Efficient
sectors and each sector is assigned with Mobile Data Cluster Centroid (GCEEC) Protocol
(MDC) node that collects the data from the CH. MDC node
computes an efficient route that is found by Expectational The Gateway Energy-Efficient Centroid (GCEEC) routing
Maximization (EM) algorithm. According to the EM algo- protocol is proposed for agriculture precision WSN to
rithm, MDC computes the route by considering the CH improve the load balancing among CHs and energy con-
residual energy and location. MDC moves to collect the data sumption of the whole network. The GCEEC protocol selects
from the CH first, whose residual energy is minimum. Simi- the efficient location of CH near the energy centroid position
larly, the MDC node collects data from other CH on an effi- and for gateway node selection for transmitting the data
cient route and delivered to the BS. towards the BS via multihop communication which maxi-
The authors in [37] proposed a Cluster Aided Multipath mizes the CH coverage and reduces the transmission power
Routing (CAMP) protocol which divided the region of inter- of CH. This section is divided into two subsection network
est into virtual zones and assign one CH for each cluster. The setup modules and process module. The network setup mod-
noncluster member’s condes have adopted the trade-off ule presents the energy consumption model, energy centroid
method for residual energy evaluation between itself and position, gateway node weight, and CH joining weight used
neighbor nodes and take decision. During this process, if in GCEEC protocol. The processing module explains the
the cluster member node is selected as the next forwarder, setup phase, transmission phase, and rotation phase of
then it cancels the trade-off method and forwards the data GCEEC.
to the CH via multihop communication. The authors
claimed that the proposed CAMP protocol improves the 3.1. Network Setup Module. The network model consists of
energy consumption due to randomly selection of CH or 100 sensor nodes and one BS. Figure 2 shows the sensor
Journal of Sensors 5

Table 1: Protocol strategies and limitations.

S# Authors Cluster Strategies Limitations

Flat routing protocols
Mobile agent route among the sensor for
Framework support small route mobile
1 ETMAM [28] 2014 ✘ data aggregation considering energy and
agent and response time is low
trust metrics
Link condition for optimize path, prevent
2 PHADOV [29] 2014 ✘ route exhausting, and reduce route error Routing overhead increase
message
Data mule (mobile node) that has the
In sufficient for different constraint and
3 OTDS [30] 2015 ✘ ability to collect and store data from
energy hole problem
sensor node and transmit towards BS
Generate trust list to avoid
4 PDORP [31] 2016 ✘ Cause significant delay
acknowledgement
Hierarchical routing protocol
Mobile sink for collecting data to alleviate Protocol not fully addressed hole problem
5 MISSEEP [32] 2015 ✓
hole due to load balancing issue
DCEM not consider the optimal location
Minimum inter cluster multihop routing of CH in cluster
6 DCEM [33] 2016 ✓
cost function Intercluster multihop routing among CH
consume more energy of CH
CH is selected by fitness functions that
consider the distance between sensor node Robustness of the algorithm, however,
7 PSO-ECHS [23] 2017 ✓ and BS, as well as sensor node and needs to be verified with the
neighbor nodes, and the residual energy of heterogeneous nature of nodes
sensor nodes
MAX-dist consume more energy of CH in
8 EECRP [34] 2017 ✓ CH selected in energy density node region
caching and transferring data
Not focus on energy-efficient optimize
9 Awan et al. [35] 2018 ✓ Cluster size reduction
route among cluster head
Mobile data cluster node utilizes as CH Subcluster nodes are taking more
10 MEACBM [36] 2019 ✓
data collection and transfer to BS processes and lead to network overhead
Adjusts the tuning factors including Has significant delay due to its energy
5 CAMP [37] 2019 ✓ remaining energy, node degree, distance calculation and randomly selection of CH
towards the sink node. in the network.

nodes which are randomly distributed in the sensor field. 3.1.2. Energy Centroid. Centroid is the mechanical term,
Each sensor node after sensing sends the data to the regional which means the imaginary central point of mass concen-
CH, then transfer the data towards the BS via single-hop tration. It is the central point where the entire mass of
direct transmission or multihop gateway nodes; it depends object is concentrated. Similarly, energy centroid in cluster
on the distance between CH and BS. is the point where sensor node is having massive energy
concentration which is distributed. Energy centroid [34]
3.1.1. Energy Consumption Model. Most of the energy is con- can be mathematically represented as in Equations (2)
sumed by the sensor node during data transmission and and (3), respectively.
receiving. The most popular and common energy model is
proposed in [34] as shown in the following:

∑ni=0 Eirs /Eo X
X ec = , ð2Þ
(
 N
l er + et + ∈fs d 2 , if d ≤ d Th ,
E=
 ð1Þ
l er + et + ∈mp d 4 , if d ≥ d Th ,

∑ni=0 Eirs /Eo Y
Y ec = , ð3Þ
N
where l is the packet size, er and et are the transmitting and
receiving energy, ∈fs and∈mp are required energy to send in where Eirs = residual energy of node i, Eo = initial energy, X
free space and multipath, respectively. The transmission and Y are the coordinate of node i, N = total number of
energy consumption depends on distance d. nodes in cluster, X ec and Y ec are the energy centroid:
6 Journal of Sensors

BS

Sensor node
Cluster head
Gateway node

Figure 2: Network topology.

1-byte 1-byte 1-byte 1-byte 2-bytes

Message type Sender’s ID X coordinate Y coordinate Energy level

Figure 3: LOCATION message.

Distance from the energy centroid position to the ith sen- adjacent to neighbor CH j, dðj, xÞ = distance between
sor node for calculating candidate CH can be shown below. adjacent CH j to cluster member node x of CH i, dðj, sÞ =
distance between CH j to BS, and dði, sÞ = distance between
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
CH j to BS:
d= ðX ec − X i Þ2 + ðY ec − Y i Þ2 : ð4Þ
Higher weightage of node becomes a cluster gateway
node.
3.1.3. Gateway Node. Information gathering from sensor
nodes and transmitting towards the BS is the main function 3.1.4. Cluster Head Joining Weight Function. When CH
of CH. Due to heavy responsibilities on CH due to the man- sends join request to neighbors, then in response, sensor
agement of cluster data, the CH consumes more energy and nodes decide to be part of cluster or not base on CH join-
sends the data directly to the BS or itself relaying on other ing weight function. The function consists the following
CH and forwards the data towards the BS. Therefore, gate- parameter, CH residual energy Eresidual ðCH j Þ, distance from
way node is formed in each cluster by CH which relay data CH to sensor node distðsi , CH j Þ, distance from CH to
towards the BS. The nodes in cluster which are adjacent to BS distðCH j , BSÞ [23].
neighbor CH are called gateway nodes. Every CH computes
the gateway node weight [38] by considering the CH residual
energy, distance between the nodes in particular cluster and


 Eresidual CH j
adjacent neighbor CH. The function is as follows: CH joining weight si , CH j =

:
dist si , CH j ∗ dist CH j , BS
  " #
SðiÞ:E 2 2 2 dð j, sÞ2 ð6Þ
Gði, jÞ = + dði, jÞ + dði, xÞ + dð j, xÞ + ,
SðiÞ:Max d ði, sÞ2
ð5Þ 3.2. Process Module. In most of the agriculture precision
WSNs, energy is the main concern due to limited resources
where SðiÞ: E = residual energy of CH, SðiÞ: Max = initial of sensor nodes. The main objective of this study is design
energy, dði, jÞ = distance between CH i and CH j, dði, xÞ = the protocol for energy-saving and efficiently utilize the
distance between CH i to cluster member node x which are resource during data processing. Clustering protocols
Journal of Sensors 7

1-byte 1-byte 2-byte

Start
Message type CH ID Avg energy

Figure 4: FEEDBACK message.

BS select CH

consist of three main phases: CH selection phase, gateway

selection phase, data transmission, and CH rotation phase.
No
3.2.1. CH Selection Phase. Initially, the BS broadcasts the Node position
HELLO-MSG across the network. Hello message contains close to energy
the BS ID and location. BS has more energy than ordinary centroid
sensor nodes, and if we use the BS to broadcast the Hello position??
messages to other sensor nodes, it will decrease the load on
Yes
other member nodes in the network. The sensor nodes send
a reply with LOCATION message as shown in Figure 3.
Message type contains the type of message. The “sender Node become CH
ID” contains the sensor node ID. The “X coordinate and Y
coordinate” are the locations of the sensor node. The energy
level contains the state of the sensor node. The LOCATION
message size is 6 bytes as shown in Figure 3. CH send joining
BS computes the average energy of the network and cal- message to
culates the energy centroid positions in the network. After neighbor nodes
the calculation of energy centroid positions, BS divides the
network into a cluster around the energy centroid position Neighbor node
and chooses the CH. The BS chooses the CH node from computes joining
the cluster which is nearest to the energy centroid posi- weight
tion. After selecting the CH, the BS broadcasts the FEED-
BACK message to the specific cluster as shown in
Figure 4. The FEEDBACK message contains the message No
Joining
type and the information of feedback message, CH’s ID, weight value
and average energy of the network. greater?
After the first CH selection by the BS, the CH transmits
joining message containing the CH ID, energy level, and Yes
Reject CH joining
location to the neighbor sensor nodes. The sensor node that
message
receives the joining message calculates the joining weight
Join CH
value of CH. If the highest CH joining weight value is
reached, then the sensor node joins the CH as a CM.
Figure 5 shows the CH selection process.
End
3.2.2. Gateway Selection Phase. After selection of CHs, each
CM who receive adjacent CH joining request computes the
gateway node weight. The gateway node weight is then sent Figure 5: CH selection flow chart.
to CH. Higher gateway node weight value is selected for a
gateway node. Gateway node then informs the adjacent begins. CM senses the data and transmits to their particular
CH by sending gateway message containing its location CH. The CH then sends towards the BS via gateway nodes
and its CH ID as shown in Figure 6 showing the gateway or directly transmits depending on threshold distance dTh .
message for requesting the adjacent CH for its gateway Just before the end of the round, each CM sends their
node. When adjacent CH gateway node receives, it then residual energy and location to CH. After that, the CH node
computes the route towards the BS via adjacent gateway calculates the average energy and energy centroid position.
node multihopping. The CM that is near to the energy centroid position will be
The data transmission of CH via gateway node depends the next candidate CH. The following conditions for the
on distance between itself and BS. If distance is less than CM can become the CH for the next round.
threshold distance ðd Th Þ, then CH sends directly to BS; oth-
erwise, CH uses gateway node for data transmission towards (i) Its energy level is greater than the average energy of
BS. Figure 7 shows the gateway selection process. cluster
3.2.3. Data Transmission and CH Rotation Phase. After the (ii) Its distance from the energy centroid position of
selection of CH and gateway node, the data communication cluster is the smallest
8 Journal of Sensors

1. Input: N=Total number of nodes

2. Output: BS divide entire network into Cluster
3. for i=1:N
4. Node transmit their ID, Location and Energy level to BS
5. end for
6. E = Average Energy of entire network calculate by BS
7. c = centroid position of cluster calculate by BS = ðX c , Y c Þ
8. for i = 1 : C
9. ci = ðX ci , Y cj Þ
10. for j = 1 : N
11. d i = distðnodeðiÞ, ci Þ
12. if (nodeðjÞ very close to ci )
13. Select as CH j
14. end if
15. end for
16. end for
17. BS Broadcast Feedback Message contain Message Type, CH ID, Avg Energy E
18. for i = 1 : N
19. nodeðiÞ receive Feedback Message f rom BS
20. if ðnodeðiÞ = = CH IDÞ
21. nodeðiÞ state = CH i
22. Broadcast joining message CH i to neighbor nodes
23. else
24. Wait for CH joining message
25. end if
26. end for
27. Neighbor nodes calculates joining weight
28. CH Joining Weight ðsi, CH j Þ = Ereidual ðCH j Þ/distðsi , CH j Þ ∗ distðCH j, BSÞ
29. if (neighbor node CH Joining Weight greater)
30. Join CH
31. else
32. Reject CH joining message
33. end if

Algorithm 1: CH Selection Phase.

4. Simulation Setup and Experimental Results

1-byte 1-byte 1-byte
Message type Gateway node ID CH ID To evaluate the performance of GCEEC with other rele-
vant algorithms, simulator selection is one of the chal-
Figure 6: Gateway message. lenge. A number of simulation environments have been
developed for wireless networks such as Objective Modular
Network Testbed in C++ (OMNET++) [39], Network
Simulator2 (NS2) [40], and MATLAB [41]. NS2 is an
Figure 8 shows the data transmission and CH rotation event-driven, open-source simulator and is the best tool
flow chart. for analyzing communication networks. The NS2 provides
The detail of GCEEC data transmission and CH rotation is executable ns script file name “Tool Command Language
discussed in Algorithm 2. Lines 1 to 8 show the data transmis- (TCL).” A simulation trace file is generated after running
sion directly or via gateway node towards the BS from the CH. TCL file which is used for plotting graphs or result analy-
From lines 8 to 11, wait till just a few moments before the end sis. NS2 provides a tool called NAM (Network AniMator)
of round, each CM sends residual energy and location to to execute the animation files.nam. NS2 comprises of two
respective CH. From line 12 to 16, it calculates the average main languages, i.e., C++ and Object-Oriented Tool com-
energy of cluster, recalculate energy centroid position, and mand language (OTcl). C++ provides user facility to
distance between each node among cluster from energy cen- describe interior working mechanisms (executed at back-
troid position. Lines 18 to 23 select the CH for next round end) of the stimulation objects, while OTcl provides the
if the energy level of node is greater than the average energy facility to setup the stimulation scripts and configuration
and distance between CH when centroid position is smaller. of objects (executed at front) [40, 42]. In addition, NS2
Algorithm 3 shows the gateway node selection process. has open-source modules and widely exploits in the
Journal of Sensors 9

Start

GW selection

Node receive No
more than one
Joining message
??

Yes

Compute GW
weight

Node remains CM
GW node No
and starts
weight
transmission
higher??
toward CH

Yes

Select as GW
node

Request adjacent
CH for GW info

No
Adjacent CH
Wait GW info
receive??

Yes

Compute route
toward BS

End

Figure 7: Gateway selection flow chart.

research community; new objects can be easily added uate the performance of proposed GCEEC protocol with
using OTcl interpreter via corresponding objects in C++ relevant scheme in terms of different performance
class. In this research work, NS2 simulator is used to eval- parameters.
10 Journal of Sensors

Start

CM periodically
send data to
CH

If dist (CH,
No BS) <
do/2

Yes
Send directly
to BS CH use GW
node to send
data toward BS

Just before No
round
over ??

Yes

If number of No
node ≥ no
of cluster

Yes

Recalculate centroid
position and Avg
energy of cluster by
CH

If energy level of CM
> Eo && dist (CM No
and energy centroid)
is smallest ??

Yes

CM status change to
new CH and prevous
CH withdraw
responsibility

GW node
selection End

Figure 8: Data transmission and CH rotation flow chart.

Journal of Sensors 11

1. do
2. CM sense and transmit sense data to CH
3. if ðdistðCH, BSÞ < d o /2Þ
4. CH directly transmit data to BS in single hope
5. else
6. CH use GW node to transmit data to BS in multi-hop
7. end if
8. while (just before round over)
9. for j = 0 : CM
10. each CM nodeðkÞ send residual energy and location to their CH
11. end for
12. CH calculate avg energy of cluster
13. Eo = ∑CMk=0 E k /CM
14. CH calculate energy centroid of cluster
15. X ec = ∑CM
i=0 ðE i rs /E o ÞX i /CM
16. Y ec = ∑CM
i=0 ðE i rs /E o ÞY i /CM
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
17. d = ðX ec − X i Þ2 + ðY ec − Y i Þ2
18. for k = 0 : CM
19. if ðenergy level of nodeðkÞ > Eo && d is smallest Þ
20. current CH change the status of node(k) = CH k
21. current CH withdraw responsibility
22. end if
23. end for
24. CH k transmit joining message as same as Algorithm 1

Algorithm 2: Data Transmission and Cluster Head Rotation.

4.1. Simulation Setup. This section presents the simulation reception of data packets. In experiments, the energy con-
setup to measure the performance of the proposed protocol sumes per round and consumes energy in clustering, assigning
design. The simulation is performed in the network, where gateway node, sensing, and transmitting of data from CM of
we set 100 ∗ 100 m area with 100 sensor nodes. All the sensor the cluster to the BS via the involvement of CH and gateway
nodes are static and know their location by means of GPS. The node. The confidence interval refers to possible range or values
BS is located outside the network at position of (100,100). We for the simulation parameters which are based on the simula-
have considered three network scenarios which are discussed tion results. The 90% confidence level is the probability that
as follow: 2%, 5%, and 10% of sensor nodes as CH. To analyze the interval contains the value of the parameter. For this study,
the performance of network by varying the number of CH in simulation confidence interval is at 90%.
sensor nodes, we ran the simulation 5 times; the average of
these instances of data is used for plotting the results. The 4.1.2. Assumptions and Limitations
simulation parameters used to evaluate the proposed protocol
with existing protocols are in Table 2. (1) Sensor nodes are static and are deployed randomly in
field
4.1.1. Performance Metrics. To determine the efficiency of the (2) All nodes adjust their transmission power according
proposed schemes against specified objectives, the following to distance
performance metrics are used.
Network lifetime. The network lifetime is the expiration (3) Communication channel is reliable and free of error
of the network life when the number of nodes depleted their (4) The sensor nodes are aware of their locations through
energy when data transmission begins; the cluster nodes some localization techniques
sense the data and send to their CH and then to BS via gate-
way node. In these experiments, the initial energy of node is (5) The BS is placed outside the network (100,100)
2 J which is reducing while transmitting and receiving control location
messages and data.
(6) Every gateway node is in the range of its neighboring
Network throughput. Network throughput refers to the
gateway node
receiving of packets by the BS. It is a successful transmission
of sensing the data from CM of clusters to the BS via the CHs 4.2. Results and Discussion
and gateway nodes.
Energy consumption. Energy consumption is the most 4.2.1. Effect on Number of Alive Nodes. The number of alive
valuable parameters for wireless sensor network in which sen- nodes which indicates the network lifetime is as shown in
sor nodes utilize their battery resources in transmission and Figures 9–11. This comparison is based on the alive nodes
12 Journal of Sensors

1. for j = 0 : CM
2. if ðnodeðjÞreceive adjacent CH joining requestÞ
3. Compute GW node Weight for Adjacent Cluster Head
4. Gði, jÞ = ½SðiÞ:E/SðiÞ:Max+½½dði, jÞ2 + dði, xÞ2 + dðj, xÞ2 + ðdðj, sÞ2 /dði, sÞ2 Þ
5. Send GW weight value to CH
6. else
7. Round Start, CM periodically send data to their CH
8. end if
9. end for
10. if ðnodeðjÞ GW weight value higherÞ
11. nodeðjÞ select as Gateway Node by CH
12. else
13. nodeðjÞ Reject as Gateway node by CH
14. end if
15. Gateway node inform its status to adjacent CH and request for Adjacent CH Gateway node
16. while (Adjacent CH Gateway Node Information Receive)
17. Compute Route
18. end while
19. if ðdistðCH, BSÞ < d Th Þ
20. CH directly transmit data to BS in single-hop
21. else
22. CH use GW node to transmit data to BS in multi-hop
23. end if

Algorithm 3: Gateway Node Selection.

die approximately at 550-800 rounds, these results are

Table 2: Simulation parameters. showing early depletion of sensor nodes. In addition, the
results also indicated that at 2% and 10% of CH, the CH
Parameter Values
consumes more energy, so that node dies earlier. While
Network area 100 m ∗ 100 m in 5%, nodes die slowly. Therefore, 5% of nodes that are
BS location At the edge of the area CHs have better conditions.
Number of sensor nodes 100
Initial energy (J) 2J
4.2.2. Average Data Transmission. Figure 12 shows the result
Data aggregation energy 5 nJ/bit/signal of average data transmission by changing the number of CH
Transmission energy 50 nJ/bit nodes in the network. As compared to EECRP, CAMP, and
Reception energy 50 nJ/bit MEACBM schemes, the proposed scheme of GCEEC per-
Data transmission rate 5000 bps forms better because CM data are transmitting towards the
ϵfs 10 pJ/bit/m2
BS via the CH and multihop gateway node. Therefore, aver-
age data transmission enhances by increasing the number
ϵ amp 0.0013 pJ/bit/m4 of CH. Furthermore, a lot of fluctuations in EECRP scheme
Round time 2 sec/round is due to the concept of MAX-dist. MAX-dist is the threshold
Packet size 200 bits distance in which the CH sends the data to the BS success-
MAC 802.11 fully if the distance is less than the data forwarded; otherwise,
CH stores the data in cache and wait for the next round due
to which average data transmission reduces. In addition,
Figure 12 also reveals when the number of CH is small, the
vs. the number of rounds at 2%, 5%, and 10%, respectively. average data transmission is less because the distance
As it is shown that the proposed protocol GCEE performs between the sensor nodes and the BS is large, and large
better as compared to EECRP, CAMP, and MEACBM. The amount of energy was consumed by all sensor nodes. As
first node dies in the proposed scheme approximately in number of CH increases, the average data transmission
between 700 and 800 rounds. Similarly, the first node dies increases. But as the number of CH increases by 5%, the aver-
in EECRP very shortly because data transmitted by CH is age data transmission also reduces because amount of data to
on a single hop, while the proposed scheme utilizes multihop increase in network which creates energy dissipation is
gateway nodes for data transmission from CH to BS. quicker among sensor nodes. The results indicate that the
Moreover, as shown in the results that the CAMP protocol proposed protocol has stable data transmission as compared
has almost the same results as MEACBM and the nodes to EECRP, CAMP, and MEACBM.
Journal of Sensors 13

120

100

Number of alive nodes

80

60

40

20

0
0 200 400 600 800 1000
Number of rounds
EECRP [31] CAMP [34]
GCEEC MEACBM [33]

Figure 9: Number of alive nodes at 2% CH.

120

100
Number of alive nodes

80

60

40

20

0
0 200 400 600 800 1000
Number of rounds
EECRP [31] CAMP [34]
GCEEC MEACBM [33]

Figure 10: Number of alive nodes at 5% CH.

4.2.3. Round vs. Packet Received by Base Station. Number of posed scheme at 5% of CH performs better transmission of
data packet received by BS which is taken into consideration packet than 2% and 10% because in 5% of CH, nodes die
at different number of rounds. Figures 13–15 show the data slowly and have better coverage and have less burden on
packet received by BS at 2%, 5%, and 10% of CH, respec- gateway nodes. While in 2% of CH, the distance between
tively. As it is shown, EECRP considerably receives fewer nodes and BS is greater so large amount of energy is con-
data packet to BS than GCEEC. Furthermore, packet received sumed. Similarly, in 10% of CH, data transmission in net-
by BS in EECRP scheme is slower than GCEEC because of work enhances; more data is relaying on gateway nodes
the adjustment of MAX-dist in EECRP where as in GCEEC, which shorten network lifetime.
gateway nodes relay data from the CH to the BS. As
compared to CAMP and MEACBM, the proposed protocol 4.2.4. Rounds vs. Energy Consumption. As shown in
GCEEC has better data transmission. However, the Figures 16–18, the total energy consumption in EECRP is
MEACBM is better than EECRP due to the use of coverage high as compared to the proposed schemes. It is due to the
and connectivity in the network by constructing a subcluster fact that EECRP scheme uses single-hop transmission by
and computing the multihop route for interclustering combi- CH as well as threshold distance name MAX-dist. The
nation among clusters and subclusters. In addition, the pro- single-hop transmission towards the BS causes load on the
14 Journal of Sensors

120

100

Number of alive nodes

80

60

40

20

0
0 200 400 600 800 1000
Number of rounds
EECRP [31] CAMP [34]
GCEEC MEACBM [33]

Figure 11: Number of alive nodes at 10% CH.

18

16
Average data transmission (KBits)

14

12

10

0
1 2 3 4 5 6 7 8 9 10
CH percentage
EECRP [31] GCEEC
CAMP [34] MEACBM [33]

Figure 12: Average data transmission with different numbers of CHs.

CH, which consumes more energy. Furthermore, due to the CH near the energy centroid position which maximizes
MAX-dist concept, i.e., CH stops transmission, stores data the network coverage of cluster nodes and reduces the energy
in cache when distance between CH and BS is greater than consumption. Furthermore, gateway nodes are selected
MAX-dist, and transmits all cache data to incoming CH among clusters which relays itself as well as other CHs and
during rotation phase. This MAX-dist cache process forward the data towards the BS which significantly reduces
consumes more energy of CH during transmission and load on CH. The experimental results indicated that GCEEC
reception. In GCEEC, the load of CH is distributed due to performs better than EECRP, CAMP, and MEACBM proto-
selection of multihop gateway node which significantly cols. All sensor nodes transmit limited amount of data to the
reduces energy consumption. Therefore, overall energy con- CH, and the CH can bear all cluster node data in his memory.
sumption is reducing in the proposed protocol as compared It can easily transfer to its gateway, and gateway can easily
to EECRP, CAMP, and MEACBM. transfer the data to the next gateway and then further trans-
The objective of these experiments is to select the CH on mit towards the BS. Therefore, there will be more transmis-
efficient location in cluster and to reduce the load on CH. The sions as compare to EECRP, CAMP, and MEACBM
proposed GCEEC protocol for agriculture precision selects protocols, but it reduces the load on CH with the help of
Journal of Sensors 15

35000

30000

Packet received by BS
25000

20000

15000

10000

5000

0
0 100 200 300 400 500 600 700
Number of rounds
EECRP [31] CAMP [34]
GCEEC MEACBM [33]

Figure 13: Packet received by BS 2% CH.

80000

70000

60000
Packet received by BS

50000

40000

30000

20000

10000

0
0 100 200 300 400 500 600 700
Number of rounds

EECRP [31] CAMP [34]

GCEEC MEACBM [33]

Figure 14: Packet received by BS 5% CH.

gateway node. Hence, we say that the proposed GCEEC pro- After designing the protocol, now we compare the whole
tocol has much more energy and efficient as compared to system performance with existing systems in agriculture
state-of-the-art routing protocols. precision field. Table 3 presents the comparison of some
The applications for precision agriculture have deployed of the existing agriculture precision systems and proposed
to analyze the environmental parameters such as humidity, system in terms of overhead, coverage area, energy consump-
crop conditions, and soil monitoring. All the data communi- tion, network lifetime, scalability, and other performance
cation process among small sensor nodes is based on feasible parameters. Table 3 indicates that most of the existing sys-
and energy-efficient sensor systems to improve the monitor- tems have more overhead and not scalable to adjust in other
ing systems for further decision making. Routing protocols agriculture fields like [43, 44]. The proposed system is scal-
are playing a very crucial role for data collection in field. able especially for agriculture precision applications such as
Complex routing protocols lead to consuming more energy, precision farming, horticulture, orchard, precision agricul-
overhead, and packet dropping. In this paper, after designing ture, precision fruticulture, precision horticulture, quality,
the proposed GCEEC routing protocol, we analyze the tree fruits, and vegetables. The table below also indicates
performance with state-of-the-art routing protocols and the different parameters to evaluate the existing agriculture
observed that proposed protocol consumes less energy precision system and their possible applications in terms of
which impacts on better data delivery in agriculture fields. network overhead, coverage area, energy consumption
16 Journal of Sensors

70000 250

60000
200
Packet received by BS

Energy consumption (J)

50000

40000 150

30000
100
20000

10000 50

0
0 100 200 300 400 500 600 700 0
0 100 200 300 400 500 600 700
Number of rounds
Number of rounds
EECRP [31] CAMP [34]
MEACBM [33] EECRP [31] CAMP [34]
GCEEC
GCEEC MEACBM [33]
Figure 15: Packet received by BS 10% CH. Figure 17: Energy consumption at 5% CH.

250 250

200 200
Energy consumption (J)

Energy consumption (J)

150 150

100 100

50
50

0
0 100 200 300 400 500 600 700 0
0 100 200 300 400 500 600 700
Number of rounds
Number of rounds
EECRP [31] CAMP [34]
MEACBM [33] EECRP [31] CAMP [34]
GCEEC
GCEEC MEACBM [33]
Figure 16: Energy consumption at 2% CH.
Figure 18: Energy consumption at 10% CH.

reduction, network lifetime, scalability, and system limita- nodes. Most of the routing protocols do not consider load
tions. Some systems have moderate but still suffered with balancing for a feasible routing path. This research improves
other parameters such as [41] which is moderate but still suf- load among the sensor nodes especially on the cluster head
fered in network lifetime. The system in [42] has high over- (CH). Furthermore, research also improves the optimized
head and also not considered energy consumption and location and rotation of CH among energy density sensor
scalability. The systems [39, 40, 44] are not scalable and also nodes. In this research, Gateway Clustering Energy-Efficient
suffered in overhead issues. The proposed system GCEEC is Centroid-based Routing Protocol (GCEEC) is proposed for
scalable and offers moderate network lifetime. WSN. The proposed protocol selects and rotates the CH near
the energy centroid position of the cluster. In addition, each
5. Conclusion CH chooses the gateway node for multihoping itself and
other CH data towards the BS which reduce load among
Wireless sensor network (WSN) is one of the emerging tech- the CH. We performed the experiment on a well-known net-
nique and technology especially for agriculture sector. In work simulator named NS-2.35 to analyze the performance
WSN, sensor nodes sense the physical and environmental of GCEEC for different criterion which includes network life-
conditions of soil and crop and send the data to the sink node time, network throughput, and energy consumption. The
by single-hop or multihop communication. Due to low com- experimental result revealed that network lifetime, through-
putational power and limited battery resources, complex put, and energy consumption of our protocol is better than
routing processes may cause energy depletion of the sensor EECRP protocol. In the future, we will analyze the proposed
 Journal of Sensors

Table 3: Comparison table.

Agriculture precision Coverage Energy consumption Network

S# Applications Overhead Scalability Limitation
system references area reduction lifetime
Time consuming,
1 [45] Greenhouse Moderate 50 m Moderate Limited Moderate
synchronization accuracy
Not feasible for other
2 [46] Agriculture High 150 m Not considered Moderate Not considered quality of services
parameters
20 times more compared to Not reliable
3 [47] Crop farming Moderate 180 m traditional systems without Limited Not considered communication beyond
CH 80 m
Its solar system has
4 [44] Tree monitoring High 180 m High Limited Not considered irregularity on different
aspects
System is more complex
5 [48] Precision agriculture Moderate 40 m High Limited Not considered and leads to predefined path
issues
Packet loss, energy
6 [43] Forest monitoring High 672 m High Moderate Not considered
consumption
Proposed system Fixed for agriculture
7 Precision agriculture Low 200 m Low Long Considered
(GCEEC) precision only
17
18 Journal of Sensors

protocol with other environments like drone-assisted WSN, [11] J. Lloret, M. Garcia, D. Bri, and J. Diaz, “A cluster-based
wireless body area networks, and sensor-based transporta- architecture to structure the topology of parallel wireless sen-
tion systems. sor networks,” Sensors, vol. 9, no. 12, pp. 10513–10544, 2009.
[12] T. Kalaivani, A. Allirani, and P. Priya, “A survey on Zigbee
Data Availability based wireless sensor networks in agriculture,” in 3rd Interna-
tional Conference on Trendz in Information Sciences & Com-
No data have to be included. puting (TISC2011), pp. 85–89, Chennai, India, December 2011.
[13] K. N. Qureshi, S. Din, G. Jeon, and F. Piccialli, “Link quality
and energy utilization based preferable next hop selection
Conflicts of Interest routing for wireless body area networks,” Computer Commu-
The authors declare that they have no conflict of interest. nications, vol. 149, pp. 382–392, 2020.
[14] S. A. Kumar and P. Ilango, “The impact of wireless sensor
network in the field of precision agriculture: a review,” Wire-
Acknowledgments less Personal Communications, vol. 98, no. 1, pp. 685–698,
2018.
This work has also been partially supported by the European
Union through the ERANETMED (Euromediterranean [15] M. H. Anisi, G. Abdul-Salaam, and A. H. Abdullah, “A survey
of wireless sensor network approaches and their energy con-
Cooperation through ERANET joint activities and beyond)
sumption for monitoring farm fields in precision agriculture,”
project ERANETMED3-227 SMARTWATIR. Precision Agriculture, vol. 16, no. 2, pp. 216–238, 2015.
[16] D. S. Long and J. D. McCallum, “On-combine, multi-sensor
References data collection for post-harvest assessment of environmental
stress in wheat,” Precision Agriculture, vol. 16, no. 5,
[1] K. Sneha, R. Kamath, M. Balachandra, and S. Prabhu, “New pp. 492–504, 2015.
gossiping protocol for routing data in sensor networks for pre-
[17] X. Fu, G. Fortino, W. Li, P. Pace, and Y. Yang, “WSNs-assisted
cision agriculture,” in Soft Computing and Signal Processing:
opportunistic network for low-latency message forwarding in
Proceeding, pp. 139–152, Springer, 2019.
sparse settings,” Future Generation Computer Systems,
[2] K. N. Qureshi and A. H. Abdullah, “Adaptation of wireless vol. 91, pp. 223–237, 2019.
sensor network in industries and their architecture, standards
[18] S. Palaniappan and P. Periyasamy, “WISEN: mote as an inno-
and applications,” World Applied Sciences Journal, vol. 30,
vative approach in precision agriculture monitoring using
no. 10, pp. 1218–1223, 2014.
wireless sensor network,” International Journal of Printing,
[3] K. N. Qureshi, A. H. Abdullah, F. Bashir, S. Iqbal, and K. M. Packaging & Allied Sciences, vol. 4, no. 5, pp. 3475–3487, 2016.
Awan, “Cluster-based data dissemination, cluster head forma-
tion under sparse, and dense traffic conditions for vehicular ad [19] A. Mehmood, S. Khan, B. Shams, and J. Lloret, “Energy-effi-
hoc networks,” International Journal of Communication Sys- cient multi-level and distance-aware clustering mechanism
tems, vol. 31, no. 8, article e3533, 2018. for WSNs,” International Journal of Communication Systems,
vol. 28, no. 5, pp. 972–989, 2015.
[4] M. Salayma, A. Al-Dubai, I. Romdhani, and Y. Nasser, “New
dynamic, reliable and energy efficient scheduling for wireless [20] N. A. Pantazis, S. A. Nikolidakis, and D. D. Vergados, “Energy-
body area networks (WBAN),” in 2017 IEEE International efficient routing protocols in wireless sensor networks: a sur-
Conference on Communications (ICC), pp. 1–6, Paris, France, vey,” IEEE Communications Surveys & Tutorials, vol. 15,
May 2017. no. 2, pp. 551–591, 2013.
[5] T. Rault, A. Bouabdallah, and Y. Challal, “Energy efficiency in [21] A. Mehmood, J. L. Mauri, M. Noman, and H. Song, “Improve-
wireless sensor networks: a top-down survey,” Computer ment of the wireless sensor network lifetime using LEACH
Networks, vol. 67, pp. 104–122, 2014. with vice-cluster head,” Ad Hoc & Sensor Wireless Networks,
[6] K. O. Flores, I. M. Butaslac, J. E. M. Gonzales, S. M. G. Dumlao, vol. 28, no. 1-2, pp. 1–17, 2015.
and R. S. J. Reyes, “Precision agriculture monitoring system [22] C. M. de Farias, L. Pirmez, G. Fortino, and A. Guerrieri, “A
using wireless sensor network and Raspberry Pi local server,” multi-sensor data fusion technique using data correlations
in 2016 IEEE Region 10 Conference (TENCON), pp. 3018– among multiple applications,” Future Generation Computer
3021, Singapore, Singapore, November 2016. Systems, vol. 92, pp. 109–118, 2019.
[7] X. Feng, J. Zhang, C. Ren, and T. Guan, “An unequal clustering [23] P. C. S. Rao, P. K. Jana, and H. Banka, “A particle swarm
algorithm concerned with time-delay for internet of things,” optimization based energy efficient cluster head selection algo-
IEEE Access, vol. 6, pp. 33895–33909, 2018. rithm for wireless sensor networks,” Wireless Networks,
[8] D. M. Omar and A. M. Khedr, “ERPLBC-CS: energy efficient vol. 23, no. 7, pp. 2005–2020, 2017.
routing protocol for load balanced clustering in wireless sensor [24] K. Guravaiah and R. L. J. A. Velusamy, “BEACH: balanced
networks,” Adhoc & Sensor Wireless Networks, vol. 42, 2018. energy and adaptive cluster head selection algorithm for wire-
[9] C. Savaglio, P. Pace, G. Aloi, A. Liotta, and G. Fortino, “Light- less sensor networks,” Ad Hoc & Sensor Wireless Networks,
weight reinforcement learning for energy efficient communi- vol. 42, 2018.
cations in wireless sensor networks,” IEEE Access, vol. 7, [25] X. Fu, G. Fortino, P. Pace, G. Aloi, and W. Li, “Environment-
pp. 29355–29364, 2019. fusion multipath routing protocol for wireless sensor net-
[10] M. Srbinovska, C. Gavrovski, V. Dimcev, A. Krkoleva, and works,” Information Fusion, vol. 53, pp. 4–19, 2020.
V. Borozan, “Environmental parameters monitoring in preci- [26] X. Liu, “Atypical hierarchical routing protocols for wireless
sion agriculture using wireless sensor networks,” Journal of sensor networks: a review,” IEEE Sensors Journal, vol. 15,
Cleaner Production, vol. 88, pp. 297–307, 2015. no. 10, pp. 5372–5383, 2015.
Journal of Sensors 19

[27] N. Jan, N. Javaid, Q. Javaid et al., “A balanced energy- networks,” IET Wireless Sensor Systems, vol. 6, no. 1,
consuming and hole-alleviating algorithm for wireless sensor pp. 17–25, 2016.
networks,” IEEE Access, vol. 5, pp. 6134–6150, 2017. [44] T. Zou, S. Lin, Q. Feng, and Y. Chen, “Energy-efficient control
[28] G. P. Gupta, M. Misra, and K. Garg, “Energy and trust aware with harvesting predictions for solar-powered wireless sensor
mobile agent migration protocol for data aggregation in wire- networks,” Sensors, vol. 16, no. 1, p. 53, 2016.
less sensor networks,” Journal of Network and Computer [45] Y. Song, J. Ma, X. Zhang, and Y. Feng, “Design of wireless sen-
Applications, vol. 41, pp. 300–311, 2014. sor network-based greenhouse environment monitoring and
[29] H. Safa, M. Karam, and B. Moussa, “PHAODV: power automatic control system,” Journal of Networks, vol. 7, no. 5,
aware heterogeneous routing protocol for MANETs,” Jour- p. 838, 2012.
nal of Network and Computer Applications, vol. 46, [46] S. Nikolidakis, D. Kandris, D. Vergados, and C. Douligeris,
pp. 60–71, 2014. “Energy efficient routing in wireless sensor networks through
[30] X. Liu, “An optimal-distance-based transmission strategy for balanced clustering,” Algorithms, vol. 6, no. 1, pp. 29–42, 2013.
lifetime maximization of wireless sensor networks,” IEEE [47] D. L. Ndzi, A. Harun, F. M. Ramli et al., “Wireless sensor net-
Sensors Journal, vol. 15, no. 6, pp. 3484–3491, 2015. work coverage measurement and planning in mixed crop
[31] G. S. Brar, S. Rani, V. Chopra, R. Malhotra, H. Song, and S. H. farming,” Computers and Electronics in Agriculture, vol. 105,
Ahmed, “Energy efficient direction-based PDORP routing pp. 83–94, 2014.
protocol for WSN,” IEEE Access, vol. 4, pp. 3182–3194, 2016.
[48] T. H. F. Khan and D. S. Kumar, “Mobile collector aided energy
[32] M. Abo-Zahhad, S. M. Ahmed, N. Sabor, and S. Sasaki, reduced (MCER) data collection in agricultural wireless sensor
“Mobile sink-based adaptive immune energy-efficient cluster- networks,” in 2016 IEEE 6th International Conference on
ing protocol for improving the lifetime and stability period of Advanced Computing (IACC), pp. 629–633, Bhimavaram,
wireless sensor networks,” IEEE Sensors Journal, vol. 15, India, February 2016.
no. 8, pp. 4576–4586, 2015.
[33] T.-T. Huynh, A.-V. Dinh-Duc, and C.-H. Tran, “Delay-
constrained energy-efficient cluster-based multi-hop routing
in wireless sensor networks,” Journal of Communications
and Networks, vol. 18, no. 4, pp. 580–588, 2016.
[34] J. Shen, A. Wang, C. Wang, P. C. K. Hung, and C. F. Lai, “An
efficient centroid-based routing protocol for energy manage-
ment in WSN-assisted IoT,” IEEE Access, vol. 5, pp. 18469–
18479, 2017.
[35] K. M. Awan, A. Ali, F. Aadil, and K. N. Qureshi, “Energy effi-
cient cluster based routing algorithm for wireless sensors net-
works,” in 2018 International Conference on Advancements
in Computational Sciences (ICACS), pp. 1–7, Lahore, Pakistan,
February 2018.
[36] K. M. Awan, P. A. Shah, K. Iqbal, S. Gillani, W. Ahmad, and
Y. Nam, “Underwater wireless sensor networks: a review of
recent issues and challenges,” Wireless Communications and
Mobile Computing, vol. 2019, Article ID 6470359, 20 pages,
2019.
[37] M. Sajwan, D. Gosain, and A. K. Sharma, “CAMP: cluster
aided multi-path routing protocol for wireless sensor net-
works,” Wireless Networks, vol. 25, no. 5, pp. 2603–2620, 2019.
[38] F. Amin and M. Zubair, “Energy-efficient clustering scheme
for multihop wireless sensor network (ECMS),” in 17th IEEE
International Multi Topic Conference 2014, pp. 131–136,
Karachi, Pakistan, December 2014.
[39] A. Varga, “OMNeT++,” in Modeling and Tools for Network
Simulation, K. Wehrle, M. Güneş, and J. Gross, Eds.,
pp. 35–59, Springer Berlin Heidelberg, Berlin, Heidelberg,
2010.
[40] T. Issariyakul and E. Hossain, Introduction to Network Simula-
tor NS2, Springer Publishing Company, Incorporated, 2010.
[41] O. Lartillot, P. Toiviainen, and T. Eerola, “A Matlab toolbox
for music information retrieval,” in Data Analysis, Machine
Learning and Applications, pp. 261–268, Springer, Berlin
Heidelberg, 2008.
[42] K. Fall and K. Varadhan, “The ns manual,” 2011, https://www
.isi.edu/nsnam/ns/ns-documentation.html.
[43] P. Mathur, R. H. Nielsen, N. R. Prasad, and R. Prasad, “Data
collection using miniature aerial vehicles in wireless sensor