=Paper=
{{Paper
|id=Vol-2786/Paper36
|storemode=property
|title=Soft Computing based Clustering Protocols in IoT for Precision and Smart Agriculture: A Survey
|pdfUrl=https://ceur-ws.org/Vol-2786/Paper36.pdf
|volume=Vol-2786
|authors=Vatan,Sandip Kumar Goyal
|dblpUrl=https://dblp.org/rec/conf/isic2/VatanG21
}}
==Soft Computing based Clustering Protocols in IoT for Precision and Smart Agriculture: A Survey==
https://ceur-ws.org/Vol-2786/Paper36.pdf
280
Soft Computing based Clustering Protocols in IoT for
Precision and Smart Agriculture: A Survey
Vatan, Sandip Kumar Goyal
Department of Computer Science and Engineering, Maharishi Markandeshwar Deemed to be University, Mullana, Ambala,
Haryana, India
Abstract
Industrial developments in the Internet of Things (IoT) have lined the way for new fields for
the use of wireless sensor network (WSN) technologies. Agricultural monitoring is a case in
point where IoT can help improve production, quality and output yield. The use of WSN and
data mining techniques will significantly improve many of the agricultural activities. One such
activity is the management of the amount of water in planted fields. In addition, during recent
years, WSN has become a more evolving field in precision farming. The key problem in the
development of WSN is the use of energy and the improvement of the life of the nodes. This
paper provides a systematic analysis of the clustering protocols based on soft computing
approaches that are used in the agricultural domain to increase WSN‟s lifetime. Classification
is carried out according to different soft computing techniques: swarm intelligence, genetic
algorithm, fuzzy logic, neural networks. The survey will then present a comparative analysis
of soft computing techniques with a focus on their goals along with their merits and
drawbacks. This survey enables the researchers to choose the suitable soft computing
technique used by clustering protocols for WSN-based precision agriculture.
Keywords
Internet of Things, Sensor Network, Swarm Intelligence, Genetic Algorithm, Precision
Farming, Neural Network, Soft Computing
1. Introduction research in this area. Over the past decade, the
conventional farming has been oriented
Everyday objects are fitted with towards the latest use of technology called
microprocessors and transmission devices in „Precision Agriculture‟. Precision farming
the Internet of Things (IoT) era, which can essentially uses wireless sensors for data
work collectively to support us turn our aggregation, irrigation control, and
environment for the better. In agriculture it is a information transmission arrangements to
favourable field where IoT devices can farmers [2]. Farmers' knowledge of well-
mitigate several problems and deliver growing field will lead to creative and accurate
favourable solutions. Agriculture uses around farming. Innovative farming techniques can be
70 per cent of the water supplies available. utilized by using IoT devices and Wireless
There is a shortage of food and water, due to Sensor Networks (WSN) in farming. The
the growing population and declining rainfall authenticity is that agriculture is nowadays
[1]. Therefore smart, precision, information storage, multi-
in recent years there has been a proliferation of storage, and data based. PA allows for greater
flexibility in crop growing and livestock
ISIC‟21: International Semantic Intelligence Conference,
February 25–27, 2021, New Delhi, India rearing. The productivity can be improved by
Mail:vatansehrawat@gmail.com(vatan),
hodce@mmumullana.org (sandip kumar goyal)
using crop monitoring technology, and costs
2020 Copyright for this paper by its authors. Use permitted
under Creative Commons License Attribution 4.0
can be minimised as more effective treatments
International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
can be applied to crops. IoT devices can be
utilized in monitoring techniques comprising
281
of nodes that communicate with the accumulated and grouped using aggregate
atmosphere using sensors to collect functions and then transmitted to the sink,
information in actual time and relay it to a thereby saving energy [5]. Data collection can
command room for further handling [3]. be seen as a basic mechanism to minimize
energy consumption and to conserve scarce
A WSN incorporates automatic sensing,
resources. A successful strategy for data
handling and wireless communication systems
accumulation would decrease the amount of
into small units called as sensor nodes. These
network traffic in WSN environments leading
self-governing sensor nodes deploy the
to substantial energy savings. We have
geographic areas randomly and closely, track
examined the clustering in the literature [6]
the ambient environment or detect events,
and are the most widely used techniques for
digitise the data collected, and route
data aggregation in WSN. Clustering the
information to the resource-rich electronics
sensor nodes is an important mechanism that
system, stated to as the base station (BS) for
enables data to be distributed in a hierarchical
additional handling and analysis [4]. However,
network by splitting the network into small
the nodes have constrained bandwidth, power,
groups. In order to allocate the management
and resource processing and limited lifespan.
tasks efficiently between the nodes, few of
The main task of the node is to sense the target
them are chosen as the head of each cluster,
phenomenon like heat, light, pressure, and
commonly referred to as the cluster heads (
then apply the data as a query response to the
CHs), and non-CH nodes called member nodes
BS or sink. For WSNs, relative to data
join their closest CH as presented in Figure 1.
transmission, the computing energy
The member nodes transmit data to the
consumption is less.
appropriate CHs, then CHs send collected data
Therefore, Data produced from neighbouring to the BS. Since there is an equal quantity of
sensors are often redundant and highly data produced by the sensor nodes, the
correlated in precision agriculture. Therefore, clustering uses the similarity between the data
instead of transmitting the data individually to and then, by combining it, decreases the
the sink each time, a large amount of network load, resulting in a further effective
redundant data is removed if data is first energy utilization. Soft computing seeks to
find precise approximation, which provides a
Figure 1: Cluster architecture
282
robust, computationally efficient and cost- conditions in WSN. Thus, in this paper lately
effective solution that saves time in the proposed soft computing techniques based
computation. Most of these methods are clustering protocols is examined which are
essentially optimistic about influenced used for precision agriculture.
biological processes and patterns of social
behaviour. With the quick growth of soft The rest of the paper is designed as below.
computing techniques over the past period [6- Section 2 presents the use of IOT in
7], clustering protocols based on particle agriculture. Section 3 discusses the
swarm optimization, ant colony optimization, clustering characteristics in WSN. In
genetic algorithms, fuzzy logic, and neural section 4, the soft computing techniques
networks have been proposed for WSN based based clustering protocols are momentarily
precision agriculture. A concept of using soft examined. Section 5 delivers a relative and
computing techniques in WSN is to deliver systematic evaluation of the studied
adaptability and robustness in relation to protocols. After All, Section 6 introduces
network breakdown, and changing wireless the conclusion and future research trends.
Figure 2: IoT in Agriculture
283
2. IoT in Agriculture The idea of IoT entire market continuum with expanded
caught attention through MIT's Auto-ID benefits that are advanced end computer,
centre and its related publications on system, and service connectivity. IoT
market exploration. IoT is basically an provides suitable results for multiple
aggregation of various devices that applications, like smart healthcare, smart
connect, sense and cooperate with their cities, protection, industrial management
interior and exterior via the embedded and agricultural traffic congestion. In the
technologies found in IoT. IoT has agricultural field a large quantity of effort
developed the megatrend for next- has been performed on
invention technology that can affect the
IoT technology to build smart farming results 3. Clustering in WSN
[4]. By exploring various problems and
challenges in agriculture, IoT has brought a The benefits of separating the networks into
huge change in the agricultural atmosphere. clusters are a) it improves scalability b) it is
Currently, with the advent of technology, it easy to manage b) it declines the quantity of
has been predicted that farmers and data to be transmitted by aggregating and
technicians can find the resolution to the summing up the data, c) it reduces the number
problems faced by farmers such as water of relay nodes d) load balancing between
deficiencies, cost control and production issues clusters e) it improves energy efficiency and f)
by using them. State-of-the-art IoT techniques it increases network longevity etc.[10].
observed all the problems and provided results
3.1 Clustering characteristics
to improve efficiency while reducing costs.
Attempts made on networks of sensors allow Typically, we describe individually clustered
us to gather data from nodes and to transmit it WSN as having three key attributes: cluster
to central servers. Data gathered the sensors properties, CH properties, and procedure
provides information on different properties for clustering.
environmental conditions for proper Cluster properties
monitoring of the entire system [8].
Monitoring ecological conditions or crop Cluster properties are separated according to
cluster requirements, such as cluster number,
production is not only an element in crop
assessment, but there are many other factors cluster volume. Below is a short description of
influencing crop production, such as field each:
managing, land and crop monitoring, The number of clusters: The quantity
unwanted object progress, wildlife attacks, and of clusters that were created may
theft etc. Furthermore, IoT offers a well- either be constant (preset) or variable.
established arrangement of restricted resources This number is flexible in the
ensuring that the greatest usage of IoT methods which randomly select the
improves effectiveness. Figure 2 indicates a CHs between the nodes.
schematic diagram displaying agricultural
Cluster size: The size of the clusters
designs that deliver easy and cost-effective
may be identical or inequitable,
connexions across individual Greenhouse,
depending on the uniform
Livestock, and Field monitoring through
dissemination of the load amongst all
secure and perfect connectivity. The figure
clusters created. The cluster
indicates that two sensor kits have been
inequality is based on the distance
introduced to monitor soil humidity,
among sensor nodes and sink.
temperature, efficiency, and air flow [9].
284
Intra-cluster communication: The become more common than
transmission inside a cluster can be integrated methods.
either precise or multi hop depending
CH selection: That clustering
on the clustering algorithm. Some
algorithm has its own method for
clustering approaches that involve
choosing a CH. But the CH selection
multi-hop transmission among the CH
systems can generally be divided into
and members when quantity of CHs is
three groups: fixed, random, and
low and volume of the clusters high.
attribute based techniques. The CHs
Inter-cluster communication: Because are chosen in preset earlier nodes are
the nodes have small-range deployed in the sector. The CHs are
transmitters, the multi-hop solution is chosen randomly in random
a suitable method. Nonetheless, strategies, and attribute-based
several WSN applications presume processes pick them based on a few of
that there is direct contact among the their attributes, such as remaining
CHs and the BS. energy and distance from the sink.
CH properties Algorithm complexity: It shows how
an algorithm converges. Several
Since the CH option is the key component of
algorithms converge based on the
each clustering algorithm, the selected CHs
requirements of the network such as
have a major impact on the efficiency. The
the amount of CHs and several
following features of the CHs are:
converge in a continuous time
Mobility: The CHs can be static or irrespective of the requirements of the
mobile. Mobile CHs can travel for a network.
restricted distance but mobile CHs'
Clustering nature: Within the
topology managing method is more
literature several clustering protocol
complex than in a stationary CHs
have recommended for WSNs. A
network.
limited number of such solutions are
Node type: Compared to the normal based on the data centric system,
nodes, the distributed CHs around the known as reactive networks. Some of
network may be rich in resources; that the solutions suggested are
is, the network supports node constructive and do not embrace the
heterogeneities or, the network may reactivity and some of them use a
be similar, and regular nodes choose mixture.
the CHs.
4. Soft Computing Techniques based
Role: The chosen CHs will play Clustering Protocols
different roles in the network,
depending on the algorithm. Those This section reviews the existing soft
are relay and fusion functions. computing techniques based clustering
protocols proposed for agriculture monitoring.
Clustering process
4.1 PSO
Method: An algorithm for clustering
can be either distributed, or PSO [11] algorithm is a simulated by the
centralized. Due to the fact that collective conduct of bird gathering, and fish
WSNs with a large amount of nodes, schooling. It contains a swarm of particles N P
disseminated approaches have in search area, in which a particle i captures a
position X i , d and a velocity Vi , d in the d th
285
dimension of global space. Throughout the transmission cost and improved location to
search area, each particle examines its solve the hot problem and to maximize the
personal best called pBesti and a global best network lifetime. The PSO-UFC therefore
elects more MCHs in region closer to the BS,
identified as gBest . After obtaining the pBesti
so that the MCHs nearer to the BS have lesser
and gBest , Pi revises its velocity and cluster sizes to save their energy for maximum
position in every iteration by utilizing equation relay traffic. Through using unequal clustering
(1) and (2) respectively. method, the PSO-UFC effectively balances
energy consumption between the nodes and
Vi , d (t 1) w Vi ,d (t ) c1r1 ( pBesti ,d X i ,d (t )) c2 r2 ( gBest X i ,d (t ))
increases the existence of the network.
(1)
4.2 Genetic Algorithm
X i , d (t 1) X i , d (t ) Vi ,d (t 1)
(2) GA is identified as an evolutionary approach,
and it replicates the mechanism of growth to
where, w signifies the inertial weight, r1 , r2
iteratively produce optimal resolution. The
signify random number and c1 , c2 show two flow diagram for GA appears in Figure 3. It
non-negative constants termed acceleration begins with the collection of randomly
factor generally set to 2.0. This procedure is produced individual population, known as
iteratively repetitive until a static amount of chromosomes based on unique information
iterations I max is achieved. about the challenge. -- chromosome is an
collection of genes containing a portion of the
Wang et al. [12] implemented a PSO-based
solution. The fitness value is assessed on the
clustering procedure with a mobile BS known
basis of the specific problem and in the next
as EPMS, to resolve the hop spot problem in
WSN. EPMS algorithm combines strategies generation the chromosomes linked with large
for the mobile BS and virtual clusters to fitness cost are selected for the reproduction
reduce delay and optimise system life. The procedure. In following step, chromosome
virtual clustering in the EPMS takes into recombination is attained using a crossover
account remaining energy and node position process to replicate different embryos.
for improved choice of CH. The mobile BS Crossover processing fuses the two parents'
transmits Hello packages to the CHs for genetic components to produce new offspring.
information group after cluster formation, and After fusion, the elected chromosomes go
the CH with high remaining energy is selected
through a mutation procedure to produce new
for data transmission within its contact range.
children by accidentally altering genes of
Results from the simulation indicate
substantial decrease in energy consumption specific chromosomes. When the population
and delay in transmission thus optimising converges so quickly, the mutation mechanism
network life. While the EPMS algorithm's restores any missing genetic values. This
fault-tolerant mechanism restores network would change the parent chromosomes with
connectivity by evaluating the broken path, it minimum fitness cost with a novel series of
may cause large overheads for communication genes generated utilizing crossover and
at the same time. mutation processes. This process is repeated
Kaur et al. [13] suggested a PSO-based till an optimal solution is attained [14].
clustering protocol with unequal and fault Rani et al. [15] Proposed an Improved
tolerance denoted as PSO-UFC in WSN. The vigorous clustering-associated approach and
protocol solves the problem of clustering and frame relay nodes (RN) to pick the most
fault tolerance for the network's long-run
desired node in the cluster. A Genetic
service. It uses unequal clustering method to
Analytics method is used for this purpose. The
stabilize energy utilization among the Master
CHs (MCHs) to resolve the hot spot issue. The simulations show that the recommended
BS manages to choose best MCHs with greater method affects the clustering algorithm
remaining energy, smaller intra-cluster Dynamic Clustering Relay Node (DCRN)
286
Fig. 4 Flowchart of Genetic Algorithm.
regarding slot consumption, throughput in engine as described in Fig. 4. The fuzzification
communication. In this study, a basic routing component represents the input to the
method based on GA is intended to optimise respective fuzzy sets and designates
the performance metric of WSNs which helps membership level to each input set defined by
to escalate both theoretical analyses and a language such as "big," "low," "medium,"
simulations for IoT applications. High "small" and "huge." If-then rules are stored by
accessibility and productivity have been the fuzzy inference engine, by which the
increased in the system based on the execution fuzzified values are mapped to linguistic
metrics of the Fitness function and the results output variables with the support of rule base.
obtained are related through simulations. It is The results achieved from the inference system
noted that proposed approach is a much better are converted by defuzzification process like
solution associated with previously utilized averaging approach, and centroid technique,
methods and in terms of network planning into the crisp values [16].
would be a novel development in IoT Pandiyaraju et al. [17] suggested a novel
applications. intelligent routing protocol to enhance the
4.3 Fuzzy Logic longevity of the network and to ensure energy
FL is a statistical method used to convey effectiveness to supply irrigation data. This
approximate human thought. Contrasting a new, smart routing protocol utilizes fuzzy
classic set theory in which results are either rules is termed terrain dependent routing. The
real or false, FL produces intermediary values inference method was utilized to take routing
according to laws of interpretation and choices. Two routing processes called Region
variables of language. A FL system contains Dependent Routing and Equalized CH
four essential parts, specifically fuzzification, Election Routing were introduced and
defuzzification, a rule base and inference compared with the framework. The
287
agricultural region in the first process is approach outperforms the recent traditional
divided into small areas of similar size called protocols.
terrains. The nodes convey the data through Mahajan et al.[19] suggested an Effective and
multi-hop to the BS. In the second stage CH is scalable protocol for distant monitoring of
chosen utilizing fuzzy rules that take into farms in rural areas, called the CL-IoT. A
account the distance to the sink and the cross-layer algorithm has been developed to
remaining energy. Ultimately, relay node minimise network transmission delay and
choice is performed utilizing Fuzzy rules. energy usage. The cross-layer dependent
Fuzzy laws are constructed using the optimal selection solution for the CH
remaining energy value, the distance from the presented to resolve the energy irregularity
BS and the head-degree. From the simulation issue. The sensor's factors of dissimilar layers
outcomes it is noted that algorithm works such as a physical, MAC, and routing layer
better in terms of lifespan, energy utilized to determine optimum CH. The
consumption than the other current algorithms. nature-inspired procedure with a new
Rajput et al. [18] recommended a clustering probabilistic choice rule acts as a fitness value
algorithm focused on fussy methods to to determine the best path for data transfer.
enhance WSN's stability. Fuzzy methods are The performance of CL-IoT is evaluated by
employed to resolve inconsistencies that exist using energy consumption, computational
in WSN. Clusters are created using the effectiveness, and QoS-effectiveness.
algorithm Fuzzy-c-means (FCM). The goal is
Yassine et al. [20] suggested a Fuzzy-based
to cluster the sensors appropriately to
clustering methodology applied to agriculture.
minimise contact gaps in the intra-cluster. It
The authors utilized FCM procedure to boost
then selects the CHs based on the FL. The
the CH selection and cluster creation scheme.
efficiency of the protocol proposed for a rise in
Each node contributing in the CH selection
reporting region and node intensity is
assesses its fitness cost, which is expressed in
observed. It can be utilized for IoT-based
terms of node intensity, and energy utilization.
WSN, due to improved network reliability and
The cluster forming process also takes
sustainability. Compared to similar recent
advantage of FCM algorithms and effectively
traditional protocols, the proposed FCM
shapes clusters. The proposed procedure is
algorithm-based clustering maximizes the
contrasted with the existing routing protocols
lifetime in the event of a rise in node intensity.
by considering measurement metrics like the
As the contact gaps inside the intra-cluster are
quantity of live nodes, network energy usage
greatly reduced. The proposed clustering
and number of data packets collected by sink.
protocol offers reliable and sustainable
The presented procedure is evaluated by
network efficiency on different scenarios. The
performing multiple simulation experiments,
simulation results indicate that regarding
and all the plotted results are based on multiple
lifetime and energy efficiency the proposed
simulation test averages. It is clearly detected
Fig. 4 Fuzzy Logic System.
288
from the results obtained that the topology From the results of the simulation, it is noted
achieves efficiently in both random and grid- that the proposed protocol improves network
based network topology. lifespan, energy consumption than the other
current algorithms.
Rajput et al. [21] effort is produced to project a
clustering algorithm to achieve balanced WSN 4.4 Artificial Neural Networks
while optimising node intensity and exposure
area. The main goal is to maximise energy A neural network is a wide interconnected
productivity by dropping the space of nodes to network of elements generated by the human
data transmission using the clustering neurons. -- neuron conducts a small amount of
algorithm FCM. The next goal is to pick an processes, and the total process is the weighted
effective CH node based on apparent number of these. A neural network must be
likelihood to achieve scalability. The outcomes trained to generate the necessary outputs by a
attained suggest that the proposed protocol is known collection of inputs. Training is
more energy effective than other related typically achieved by feeding patterns of
approaches. Thus it can be used effectively in teaching into the system and letting the
IoT systems for farm monitoring. FCM network change its weighting role according to
algorithm is implemented in order to shape some previously established rules of learning.
optimum network clusters with Node positions The learning may be supervised or
and cluster number are the inputs. It unmonitored. An ANN essentially comprises
approximates cluster centre points for better of three layers: input, hidden layer, and output,
cluster structure in surveillance property. This where each layer can have numbers of nodes
greatly decreases the data communication in it. This tests the output of the neural
distances between the nodes. The results network against the desired output, and if the
demonstration that the protocol can shape results are not as planned, the weights among
larger clusters to maximize the WSN lifetime. layers are adapted and the procedure is
repeated until a very minor error remains [23].
Pandiyaraju et al. [22] suggested a new
intelligent routing algorithm to expand the Thangaramya et al. [24] proposed a new IoT-
network lifetime and to deliver energy based sensor network routing algorithm that
efficient routing. This new smart routing uses a clustering approach based on neuro-
protocol utilizes fuzzy rules precision farming. fuzzy rule to perform cluster-based routing to
This algorithm operates in three stages, improve network presentation. In this method,
namely the phase of terrain forming, the phase the cluster forming in WSNs used the energy
of election of Terrain Head and the phase of modelling to proficiently route the packets by
routing based on terrain. The farmland in the applying machine learning using
first procedure is divided into small parts of convolutionary NN with fuzzy weight
similar size known as terrains. In the next adjustment laws, thus extending the lifetime of
phase CH is chosen utilizing fuzzy rules that the network. The authors also considered four
take into account the distance to the sink and elements, namely the remaining energy, the
the remaining energy. The TH which functions distance among the CH and BS, the distance
as a relay transmits information from source to among the node and the CH, and the CH-
sink. The relay node is chosen taking into degree, that are essential features for the usage
account the residual resources, the departure of energy and the lifetime of the network.
from the sink and the node-degree. But lastly, They tested the proposed algorithm using
in the third step, relay node selection is MATLAB simulations. The NFIS output value
performed utilizing fuzzy rules. The laws are was utilized to establish the CH for the
constructed utilizing the residual energy, the member node to meet. It has been observed
distance from the BS and the head-degree. that the proposed protocol performed better
289
regarding energy consumption and device life of luciferin is correlated with the health of the
span because of the usage of neuro-fuzzy actual location of the agent. The lighter
rules. individual means better positioning (is a better
4.5 Harmony Search Algorithm solution). Using a probabilistic process, each
agent can only be attracted within the local-
Harmony Search (HS) is an important decision domain by a neighbour whose
evolutionary approach intended to imitate the luciferin intensity is higher than his own, and
process of jazz musicians improvising. HS then shift toward it [27].
creates random resolutions that are called
Harmony Memory. In each iteration a new Li et al. [28] proposed a new paradigm of
solution is created and compared with the cluster head selection to optimise network
worst resolution. It is substituted with the lifespan and energy consumption. In addition,
severest resolution if new resolution is better. this paper suggests a Glowworm swarm based
Phase continues until condition of termination on Fitness with Fruitfly Algorithm (FGF),
is fulfilled. HS algorithm's strength lies in its which is the hybridization of GSO and Fruitfly
capability to effectively arrive at global Optimization Algorithm to pick the best CH in
solution. [25]. WSN. Despite this, GSO should deal with
non-linear problems; the model fails in solving
Bongale et al. [26] Two stage Chief Election problems of large dimensions. The algorithm
Plan for the cluster. The quest algorithm for also reduces the processing speed and has
harmony in the first stage is utilized to limited local search capabilities. Similarly, in
evaluate energy efficient CHs which are search space the FFOA algorithm also has
adequately divided by some ideal distance drawbacks including lower convergence rate.
from each other. The tentatively chosen CH Therefore it hybridises the best attributes of
are then enhanced by the firefly algorithm, given algorithms to solve all the problems of
taking the parameters like node intensity, the traditional algorithm. The proposed system
cluster density and energy consumption. will minimise the negative search ability,
During the CH election method, all fireflies while the enhanced search functionality can be
undergo a recursive update and estimation used for convergence. Consequently, the
process before the CH are finalised. The various goals were successfully addressed as
approach proposed guarantees that chosen compared with the conventional algorithms.
CHs have high remaining energy and can
consume less energy in clustering. It also 4.7 Gravitational search algorithm
guarantees that clusters designed to have large This approach is stimulated by Newton's laws
intensity, minimal transmission costs and of movement and does not need a derivative to
chosen CHs are well disconnected from each find fitness function solutions. The non-
other so that CHs include the whole sensing convex fitness function can be extended to
area. The advantage of the proposed method of random search algorithms which are
cluster creation is that the energy utilization is differentiable. Masses are the agent in the
spread through the various network CHs. GSA which can explore the feasible region.
4.6 Glowworm Swarm Optimization Each agent shows a solution to the problem of
optimisation where the value of each mass
In GSO, a colony of glowworms is initially shows the modality of that solution assessed
distributed in solution space at random. -- by the function of fitness. Thus agents with
glowworm in search space represents a higher mass values provide better solutions;
solution of objective function and carries along they can entice other masses by gravitational
with it a certain amount of luciferin. The level force. Thus, the agents' global movement is
towards a better solution [29].
290
Dhumane et al. [30] suggested a multi- systems undergo the operations of crossover
objective fractional GSA to locate the and mutation to generate stable offsprings. The
optimum CH for IoT. The Fractional GSA clusters achieved from proposed GWO-GA is
(FGSA) is intended to locate the optimal CH stable in terms of balancing network load, and
in the IoT network in an iterative way to energy efficiency.
prolong the node's lifetime. In FGSA, the CH
Arikumar et al. [32] proposed an Energy
node is elected which is estimated by the
Efficient LifeTime Maximization (EELTM)
fitness value utilizing numerous goals like
methodology that uses PSO and FIS. PSO-
distance, linking duration and residual energy,
based fitness method to assess the fitness
called the multi-objective FGSA (MOFGSA).
function of every node, based on the main
The proposed technique comprises of two
energy, intensity and distance factors.
important characteristics, (1) the
Additionally, an efficient CH – CR (cluster
implementation of a MOFGSA algorithm and
router) selection algorithm that utilizes the
(2) the creation of a novel fitness model with
fitness functions determined by PSO is
several goals to prolong the lifespan. The IoT
proposed to decide two optimum sensor nodes
network is initially shaped together with one
in every cluster to serve as CH and CR. The
or more sink nodes, with the greater quantity
elected CH collects the data entirely from its
of nodes. The FGSA algorithm contains both
associated members, while the CR is
the fractional principle and GSA which is used
responsible for collecting and transmitting the
to approximate agent power, velocity, and
data collected from its CH to the sink. So CH's
position. Then the fractional principle is
operating cost is lowered. Additional smart
understood with GSA to change the location of
strategy is that FIS can evaluate the radius for
the agent to optimally figure out CH. Thus, in
CH and divide the network into unbalanced
the proposed FGSA algorithm, the four
clusters. To evaluate the ability of EELTM,
multiple targets such as distance, latency and
parameters like residual-energy, FND and 50%
connect lifetime are used to measure novel
node die are used. Thus, the EELTM method
fitness cost. The MOFGSA thus ensures the
enlarges the network lifetime.
lifespan of the IoT nodes is extended.
Robinson et al. [33] proposed an Energy
4.7 Hybrid
Aware Clustering utilizing Neuro-fuzzy
Lipare et al. [31] combined two bio-inspired method (EACNF) to generate energy efficient
techniques specifically Grey Wolf clusters in the network. The proposed structure
Optimization (GWO) and GA for balancing comprises of a fuzzy and neural network
traffic load in agriculture-WSN. There is which attains energy effectiveness in cluster
progress in the initial population process of formation and CH election. EACNF utilized
both algorithms. The crossover and mutation the neural network to deliver an appropriate
process has been improved to achieve the safe training set for the energy and node density of
off-spring to improve the network's balanced each sensor node in order to evaluation the
charge and effective energy usage. For energy consumed by Unknown CHs. The
crossover operation, the authors substituted nodes with higher residual energy are qualified
only the worst solution instead of switching all to pick energy conscious cluster heads with
the parental solutions. This improves to save different base station location. Fuzzy if – then
population's best option. In the mutation mapping rule is utilized to form clusters and
process we modified the task of the node CHs in the fuzzy logic component which
associated with the remote gateway to their inputs. For WSN, EACNF is designed to
closest gateway instead of varying a random manage Confidence factor for network
bit. This cuts the sensor node's energy security. EACNF used three metrics like the
consumption. GWO and GA's best suited range of transmission, the residual energy, and
291
the Confidence aspect to enhance network Since the great potential of IoT in all phases of
lifetime. the latest life has been broadly recognized
sensors, wireless networks and software
4. Comparative analysis
applications have become an essential and
This section delivers a comparative study of significant asset of modern agricultural
several soft computing techniques based structure. This paper surveys soft computing
clustering protocols in IoT-WSN. Table 1 techniques based clustering protocols
emphasizes the soft computing technique, employed in agriculture domain in WSN. The
clustering parameters, and transmission mode importance of soft computing techniques in
of the reviewed clustering protocols. This WSN reflects a thorough study of various
survey will also enable the researchers to clustering schemes with focus on their goals.
rapidly analyse several soft computing The comparative analysis enables the selection
techniques reviewed in this paper and elect the of suitable soft computing technique based
suitable soft computing technique based on clustering schemes used in WSN to enable
their merits and limitations as given in Table energy-efficient aggregation in smart
2. agriculture system.
5. Conclusion
Table 1: Comparative Analysis of Soft Computing Techniques Based Clustering Protocols
Protocols Soft Computing Technique Clustering Parameter Transmission type
EPMS [12] PSO Residual energy, Location Single hop
PSO-UFC [13] PSO Residual energy, Intra-cluster Multi-hop
distance, Inter-cluster distance
DCRN-GA [15] GA Residual energy, Location, Multi-hop
Distance to BS
Pandiyaraju et al. [17] Fuzzy logic Distance to BS, Residual Multi-hop
energy
Rajput et al. [18] Fuzzy logic Coverage area, Node density Single hop
CL-IOT [19] Fuzzy logic Residual energy, distance to Multi-hop
BS
Yassine et al. [20] Fuzzy logic Node density, Cluster fairness, Multi-hop
Expected energy consumption
Rajput et al. [21] Fuzzy logic Node location, Number of Single hop
clusters
Pandivaraju et al. [22] Fuzzy logic Distance to the BS, Residual Multi-hop
energy
Thangaranga et al. [24] Neural networks Residual energy, distance Single hop
among CH and the BS,
Distance among sensor node
and CH
292
Bongale et al. [26] HSA Node density, Cluster Single hop
compactness, Energy
Le et al. [28] Glowworm search Inter-cluster distance, Intra- Single hop
optimization cluster distance, Energy
MOFGSA [30] GSA Distance, Link lifetime, Single hop
Energy
Lipare et al. [31] Grey wolf optimization, Energy, Intra-cluster Distance, Single hop
Inter-cluster distance
GA
EELTM [32] PSO, Fuzzy Logic Energy, Distance, Density Multi-hop
EACNF [33] Neural network, fuzzy Energy, Density Single hop
logic
293
Table 2 Summary of soft computing techniques
Soft Merits Limitations
Computing
Approaches
Fuzzy Logic The implementation of FL needs minimum growth rate,
and design period. Fuzzy laws are not flexible about the network changing
The clustering protocols based on FL have the ability to aspects and needed to be re-learnt with vigorous
deal with contradictory conditions without needing any requirements.
complex statistical model.
Particle PSO is used due to its simplicity to implement on The iterative factor of PSO does not make it appropriate
swarm software, and extremely optimum resolution. for multimedia applications.
Optimization PSO based clustering algorithms indicate considerable PSO has significant memory limitation which involves
progress in terms of strength and flexibility resource-rich BS.
Genetic GA-based clustering proficiently resolve multi-objective GA has sluggish convergence rate which limits its
Algorithm problem where data regarding the network like network realization in multimedia applications.
topology, intensity, and dimension is not essential. It does not have the capability to contract with dynamic
system topology and transmission connection
breakdowns.
Harmony HS algorithm has relatively less parameters to calculate For complex issues, HS is not effective in detecting the
search the fitness value global solution in large search space.
It has the capability to detect areas with improved
results.
Neural NNs can contract with inadequate data sets. Unnecessary training may be needed in convoluted
Networks NNs are effective in forecast. ANN systems
Glow worm GSO does not utilize velocities, and hence has no GSO has not applied for high dimensional challenges.
Swarm difficulty as that related with velocity in PSO.
Optimization The rate of convergence is maximum in possibility of
discovering the global optimized response.
Gravitational It involves only two parameters to adapt i.e. mass & GSA has large computational cost.
Search velocity to achieve near global optimal solution. GSA has convergence issue if preliminary population
Algorithm not created well
294
Wireless Personal Communications, 107(1),
References pp.471-512.
10. Afsar, M. M., & Tayarani-N, M. H. (2014).
1. Muangprathub, J., Boonnam, N., Clustering in sensor networks: A literature
Kajornkasirat, S., Lekbangpong, N., survey. Journal of Network and Computer
Wanichsombat, A. and Nillaor, P., 2019. IoT Applications, 46, 198-226.
and agriculture data analysis for smart farm. 11. Poli, Riccardo, James Kennedy, and Tim
Computers and electronics in agriculture, 156, Blackwell. "Particle swarm optimization."
pp.467-474. Swarm intelligence 1, no. 1 (2007): 33-57
2. Sadowski, S. and Spachos, P., 2020. Wireless 12. Wang, Jin, Yiquan Cao, Bin Li, Hye-jin Kim,
technologies for smart agricultural monitoring and Sungyoung Lee. "Particle swarm
using internet of things devices with energy optimization based clustering algorithm with
harvesting capabilities. Computers and mobile sink for WSNs." Future Generation
Electronics in Agriculture, 172, p.105338. Computer Systems 76 (2017): 452-457.
3. Sanjeevi, P., Prasanna, S., Siva Kumar, B., 13. Kaur, T. and Kumar, D., 2018. Particle swarm
Gunasekaran, G., Alagiri, I. and Vijay Anand, optimization-based unequal and fault tolerant
R., 2020. Precision agriculture and farming clustering protocol for wireless sensor
using Internet of Things based on wireless networks. IEEE Sensors Journal, 18(11),
sensor network. Transactions on Emerging pp.4614-4622.
Telecommunications Technologies, p.e3978. 14. Mirjalili, S., 2019. Genetic algorithm. In
4. V. Bhandary, A. Malik, and S Kumar, Evolutionary algorithms and neural networks
“Routing in Wireless Multimedia Sensor (pp. 43-55). Springer, Cham.
Networks: A Survey of Existing Protocols and 15. Rani, S., Ahmed, S.H. and Rastogi, R., 2019.
Open Research Issues,” Journal of Dynamic clustering approach based on
Engineering, pp. 1-27, 2016. wireless sensor networks genetic algorithm for
5. J. Yick, B. Mukherjee, and D. Ghosal, IoT applications. Wireless Networks, pp.1-10.
“Wireless sensor network survey,” Computer 16. Barros, L.C.D., Bassanezi, R.C. and Lodwick,
networks, Vol.. 52, no 12, pp. 2292-2330, W.A., 2017. A first course in fuzzy logic,
2008. fuzzy dynamical systems, and biomathematics:
6. Dhasian, H.R. and Balasubramanian, P., 2013. theory and applications.
Survey of data aggregation techniques using 17. Pandiyaraju, V., Logambigai, R., Ganapathy,
soft computing in wireless sensor networks. S. and Kannan, A., 2020. An Energy Efficient
IET Information Security, 7(4), pp.336-342. Routing Algorithm for WSNs Using
7. Sharma, R., Vashisht, V. and Singh, U., 2020. Intelligent Fuzzy Rules in Precision
Soft Computing Paradigms Based Clustering Agriculture. Wireless Personal
in Wireless Sensor Networks: A Survey. In Communications, pp.1-17.
Advances in Data Sciences, Security and 18. Rajput, A. and Kumaravelu, V.B., 2020. FCM
Applications (pp. 133-159). Springer, clustering and FLS based CH selection to
Singapore. enhance sustainability of wireless sensor
8. Farooq, M.S., Riaz, S., Abid, A., Abid, K. and networks for environmental monitoring
Naeem, M.A., 2019. A Survey on the Role of applications. Journal of Ambient Intelligence
IoT in Agriculture for the Implementation of and Humanized Computing, pp.1-21.
Smart Farming. IEEE Access, 7, pp.156237- 19. Mahajan, H.B., Badarla, A. and Junnarkar,
156271. A.A., 2020. CL-IoT: cross-layer Internet of
9. Thakur, D., Kumar, Y., Kumar, A. and Singh, Things protocol for intelligent manufacturing
P.K., 2019. Applicability of wireless sensor of smart farming. Journal of Ambient
networks in precision agriculture: A review. Intelligence and Humanized Computing, pp.1-
15.
295
20. Yassine, S. and Fatima, L., 2019, October. Algorithms. Wireless Personal
Dynamic Cluster Head Selection Method for Communications, 106(2), pp.275-306.
Wireless Sensor Network for Agricultural 27. Krishnanand, K.N. and Ghose, D., 2009.
Application of Internet of Things based Fuzzy Glowworm swarm optimization for searching
C-means Clustering Algorithm. In 2019 7th higher dimensional spaces. In Innovations in
Mediterranean Congress of Swarm Intelligence (pp. 61-75). Springer,
Telecommunications (CMT) (pp. 1-9). IEEE. Berlin, Heidelberg.
21. Rajput, A. and Kumaravelu, V.B., 2019. 28. Li, Z. and Huang, X., 2016. Glowworm swarm
Scalable and sustainable wireless sensor optimization and its application to blind signal
networks for agricultural application of separation. Mathematical Problems in
Internet of things using fuzzy c-means Engineering, 2016.
algorithm. Sustainable Computing: Informatics 29. Rashedi, E., Nezamabadi-Pour, H. and
and Systems, 22, pp.62-74. Saryazdi, S., 2009. GSA: a gravitational search
22. Pandiyaraju, V., Logambigai, R., Ganapathy, algorithm. Information sciences, 179(13),
S. and Kannan, A., 2020. An Energy Efficient pp.2232-2248.
Routing Algorithm for WSNs Using 30. Dhumane, A.V. and Prasad, R.S., 2019. Multi-
Intelligent Fuzzy Rules in Precision objective fractional gravitational search
Agriculture. Wireless Personal algorithm for energy efficient routing in IoT.
Communications, pp.1-17. Wireless networks, 25(1), pp.399-413.
23. Goldberg, Y., 2017. Neural network methods 31. Lipare, A., Edla, D.R., Cheruku, R. and
for natural language processing. Synthesis Tripathi, D., 2020. GWO-GA Based Load
Lectures on Human Language Technologies, Balanced and Energy Efficient Clustering
10(1), pp.1-309. Approach for WSN. In Smart Trends in
24. Thangaramya, K., Kulothungan, K., Computing and Communications (pp. 287-
Logambigai, R., Selvi, M., Ganapathy, S. and 295). Springer, Singapore.
Kannan, A., 2019. Energy aware cluster and 32. Arikumar, K.S., Natarajan, V. and Satapathy,
neuro-fuzzy based routing algorithm for S.C., 2020. EELTM: An Energy Efficient
wireless sensor networks in IoT. Computer LifeTime Maximization Approach for WSN
Networks, 151, pp.211-223. by PSO and Fuzzy-Based Unequal Clustering.
25. Lee, K.S. and Geem, Z.W., 2004. A new Arabian Journal for Science and Engineering,
structural optimization method based on the pp.1-16.
harmony search algorithm. Computers & 33. Robinson, Y.H., Julie, E.G., Balaji, S. and
structures, 82(9-10), pp.781-798. Ayyasamy, A., 2017. Energy aware clustering
26. Bongale, A.M., Nirmala, C.R. and Bongale, scheme in wireless sensor network using
A.M., 2019. Hybrid Cluster Head Election for neuro-fuzzy approach. Wireless Personal
WSN Based on Firefly and Harmony Search Communications, 95(2), pp.703-721.