IRJET- Efficient Data Dissemination using GWO Fuzzy in Neural Networks

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 07 Issue: 04 | Apr 2020 www.irjet.net p-ISSN: 2395-0072
© 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 223
EFFICIENT DATA DISSEMINATION USING GWO FUZZY IN NEURAL
NETWORKS
Sivaganapathi.M1, Paramaiyappan.M2, Stella Rosemalar.P3, Partha Sarathi.S4
1Assistant Professor, JP College of Engineering
2Assistant Professor, JP College of Engineering
3Assosiate Professor, JP College of Engineering
4Assistant Professor, Dhanalakshmi College of Engineering
----------------------------------------------------------------------***---------------------------------------------------------------------
Abstract— The dispersed nature and vibrant topology of
WSN have some basic requirements that include reduced
energy utilization and extended network’s life time. We have
focused on hierarchical protocols. In such protocols the nodes
are arranged in clusters .To synchronize action and route
data, cluster head are selected one per cluster. We have
introduced a new approach in WSN for selecting the cluster
head by making use of GWO (Grey Wolf Optimization)
algorithm in order to increase network’s lifetime. We have
used residual energy as a factor to make cluster-head. The
simulation result provide network’s performance on the basis
of some factors including number of dead nodes, total energy
consumption and the number of packets transferred to base
station (BS) and cluster head(CH).Andanotherfactorincluded
is the Artificial Neural Network (AFF) that has been used to
predict the cluster head .The performance of proposed
algorithm is compared with LEACH and LEACH-C based on
energy efficiency and improved network lifetime. Simulation
results show that the GWO Fuzzy and neural network can
efficiently disseminate datawithahigh datadeliveryratio and
a minimized overhea.
Key Words----GWO (Grey Wolf Optimization) algorithm,
Artificial Neural Network (ANN), residual energy factor,
cluster-head, number of dead node.
1. INTRODUCTION
Wireless Sensor Network (WSN) consists of many sensor
nodes distributed randomly in a geographical area. The
nodes are placed in remote and unattended environment.
Further, the nodes are powered by batteries. Hence,
replacement or recharging of batteries is a difficult task. So,
energy efficient algorithms are suggested to reduce the
discharge rate of batteries thereby increasing the lifetime of
the node, in turn the network lifetime. Clustering isonesuch
technique employed universallytoachieveenergyefficiency.
Clustering involvestwophasesnamelyclusterformation and
Cluster Head (CH) selection. Several works focus on cluster
formation and CH selection. At this juncture, cluster
formation is grouping of nodes which are located at single
hop whereas CH selection is the process of selecting a head
from the available pool of eligible nodes either randomly or
based on certain factors. This leads to hierarchical network
with member nodes and CH as portrayed. The member
nodes send data to the CH where the data is aggregated and
further transmitted to another CH or Base Station (BS).
Hence, the energy consumed by the member node is
relatively lesser when compared with that of CH. Especially,
when the CH communicates for a long duration; its energy
drainage is high resulting in partitioning of network due to
the key node’s (CH‟s) death. To overcome this, Cluster Head
Reselection (CHR) is mandate.
Further, CH reselection may be either periodic or
threshold based. In periodic CHR, the node continues to be
CH for a fixed interval. In such case,frequentreselection may
occur when the interval is short leading to time overhead
and energy consumption.Similarly,whentheinterval is long,
it may result in network partition. Whereas in threshold
based reselection, the interval depends upon the rate of
energy drainage and threshold energy. So, to minimize
frequent CH reselection, all the major energy drainage
factors are to be considered for CH selection. Further, CH
selection may be based on fuzzy logic, neural networks and
other soft computing techniques due to its ability to resolve
uncertainties. Fuzzy logic is widely used in prediction of CH.
Fuzzy based CH selection considers the major energy
drainage factors such as Residual Power of Sensor Nodes
(RPSN), Distance of the Node from Base Station (DNBS),
Distance of the node from Cluster Centric (DCC), Degree of
Neighboring Nodes (DNN), Sensor Node Movement (SNM),
Rate of recurrent CommunicationofSensorNode(RCSN) etc.
to select the CH. Some of the proposals arevalidatedthrough
hardware. Nonetheless, in the above works thetimeinterval
between two consecutiveCH reselections,i.e.,thereselection
interval or number of rounds a node can act as CH is not
anticipated. CH selection based on energy and distance are
proposed. In prediction of reselection interval or number of
rounds of CH based on residual energy and the distance of
the node from BS is proposed. Moreover, the accuracy of
reselection interval prediction relies on the inherent energy
factors considered for prediction. However, the energy
consumed by a node with the influence of an obstacle is not
dealt.
The presence of an obstacle can be realizedthrough
Received Signal StrengthIndicator(RSSI).Ontheotherhand,
the earlier works considered RSSI toestimateeitherlocation
or path loss to analyze link quality. Also, RSSI is used as a
metric in security of WSN .Nevertheless, none of the works
considered RSSI for CH selection.

Furthermore, first order radio model is used to predict
energy consumption of the node. Nonetheless, the model
shows deviation in energy consumption measurement
through hardware .This is due to the fact that the model fails
to consider the deviation in the energy consumption by
electronic circuitry. This energyconsumptionisbasedonthe
rating and characteristics of Radio Frequency transceiver
and is product dependent. Therefore, a simple and more
reliable energy prediction model is to be developed
irrespective of the hardware used.
To address these issues, the following contributions are
made.
(i)RSSI along with residual energy and distance of the node
from Base Station are considered for CH selection. By doing
so, the prediction on the expected number of rounds for a
node to continue as CH would be more precise.
(ii) In addition, RSSI based energy prediction model using
linear regression is proposed which has significance where
practical measurement of energy is not possible.
2. RELATED WORK IN VANET
In recent years, data disseminationhasbeenheavily
researched in the context of vehicular networks. Previous
Existing work either consider the highway scenario or the
urban layout, however, efficient protocols must
accommodate both traffic scenarios. The majorityofexisting
protocols with different layout considerations aim at
addressing the broadcast storm problem for an improved
network performance in termsofcoverage,data redundancy
and dissemination delay. This problem isoftenaddressedby
selecting a set of vehicles as forwarding relays. Different
methodologies have been utilized recently to offer efficient
solutions mostly by focusing either on data redundancy or
dissemination delay. In this section, we firstly review
existing data dissemination protocols forhighwayscenarios,
and secondly for urban environments. Thirdly, we discuss
the major differences among the reviewed protocols from a
set of perspectives correlated with target scenarios, prior
assumptions, and performance evaluation.
A. Data Dissemination Protocols for Highway Scenarios
In an enhanced 1-persistance data dissemination
protocol (E1PD) is introduced for vehicular networks in
highway environments. E1PD enhances the classical slotted
1-persistance method, by making variations in the waiting
time among receiving vehicles based on their direction.This
way, the number of vehicles assigned to a singletimeslotcan
be reduced because vehicles at nearly the same location but
different direction are assigned to different timeslots.
Nevertheless, the total number of timeslots is a
predetermined constant that is not adapted to the road
traffic condition. Thus, data redundancy can easily lead to
high communication overhead due to unnecessary
transmissions under high densities. Another data
dissemination protocol for highways named ATENA is
proposed in. It selects vehicles inside a preference zone to
rebroadcast messages. Similar to, performanceresults show
that ATENA suffers from high data redundancy especially
under high-density scenarios, since many vehicles may
unnecessarily participate in relaying traffic data. The idea of
considering a preference zone to select forwarding vehicles
is also presentedinADDprotocol,whichcombinesbroadcast
suppression with delay de-synchronization. Despite the
reduced collisions, ADD does not show consistent data
delivery. In addition, existing protocols can achieve better
performance in terms of dissemination delay. Besides the
utilization of a preference zone, exploiting beacons iswidely
considered to decrease thepercentageofdata redundancy in
vehicular environments. In ADDHV, beacons are used solely
for detecting neighbors, and the propagation efficiency is
defined for further transmission control. However, and
similar to ADD, ADDHV does not show an improved
performance in terms of dissemination delay.
B. Data Dissemination Protocols for Urban Scenarios
The majority of dissemination protocols designed
for urban vehicular networks rely on the utilization if
beacons for an improved broadcast. Nakorn and
Rojviboonchai propose a density-aware reliable broadcast
protocol (DECA) for urban vehicularnetworkswithadaptive
beacon interval. Beaconing in DECA allows nodes to
exchange their local density and identifiers of received
messages with 1-hop neighbors. When a broadcast is
initiated, the neighbor with the highest local density is
selected as the next relay. To improve reliability, other
neighbors store the message and set a waiting timer, so that
another neighbor can forward the data in the case of relay
failure. Although DECA is evaluated underbothhighwayand
urban scenarios, it does not consider data dissemination in
different directions depending on the road layout.
ERD is another protocol which exploits beacons for
relay node selection to improve urban vehicular broadcast.
However, unlike DECA, ERD employs directional broadcast
by considering three road layouts: straight road, curve road,
and intersection. SimulationresultsshowthatERDimproves
bandwidth utilization while maintaining data delivery ratio.
Beacons are also utilized in TURBO to determine the status
of traffic regime. A high number of beacons indicate dense
traffic, while low or no beacons indicate a sparse network.
Each vehicle can decide to relay messages based on the
number of beacon messages received. When dense traffic is
encountered, only part of the vehicles would act as relays to
avoid the broadcast storm. Similarly,TrADprotocol requires
beaconing to maintain the status ina one-hopneighborhood.
The broadcast suppression in TrAD is senders oriented,
where the decision is made by the sender to control the
rebroadcast order of neighbors. A cluster classification
mechanism is employed to identify vehicles belonging to
each cluster.
DRIVE, AMD, and U-HyDi are three of the recent
data dissemination protocols that consider both highway

and urban traffic scenarios. Similar to they rely on local one-
hop neighborhood information via beaconing to deliver
messages under different traffic conditions. Simulation
results show that these protocols can decrease data
dissemination delay and overhead while maintaining a high
data delivery ratio. Despite the fact that the involvement of
beacons can support traffic density estimation (which is
essential to determine traffic condition for efficient data
dissemination), it has been proven that beaconing with a
fixed period have several drawbacks on the performance of
the network, such as wasted bandwidth and increased
network congestion. It was shown that channel would be
80% loaded when each vehicle sends a 200-byte beacon
every 100 milliseconds at the range of 300m. Therefore, it is
crucial for data dissemination protocols to maintain the size
of exploited beacons, in order to preserve the limited
available bandwidth. Methods that utilize beacon-free
approaches also exist. However, these methods often make
assumptions on a fixed infrastructure. For instance Li et al.
Propose the Efficient Directional Broadcast (EDB) protocol
which assumes a directional repeatertooperateateachroad
intersection. A repeater is equipped with four fixed
directional antennas pointingtofourroadsegments,inorder
to forward data messages to vehicles on different road
segments incident to an intersection. Other examples of
protocols with infrastructural assumptions arefoundin.Ina
scheduling framework is proposed to address the collision
problem. To improve dissemination efficiency, a relay
selection strategy is employed and space-time network
coding is adopted with low detection complexity and space-
time diversity. RSU is the resource that broadcasts data to
the vehicles within its coverage, and vehicles can share data
with their neighbors. Despite its proven delay performance,
the proposed strategy does not operate in the absence of
fixed infrastructure. In addition,thedisseminationoverhead
is not considered in its evaluation.
To reduce the communication overhead and
enhance data dissemination throughput in VANETs, Zeng et
al. Propose a channel prediction based scheduling strategy,
which achieves high scheduling efficiency under both urban
and highway scenarios. However, and similar to and , the
proposed strategy assumes an existing infrastructurewitha
set of RSUs that are connected to a control server through a
wired backhaul. Eachofthemcollects andmanagesvehicular
information within its communication coverage.
However, most of them suffer from an increasing
delay or high redundancy overhead. For example, UGAD
shows an efficient performance when setting appropriate
threshold values; however, data redundancy may be
unnecessarily increased under high-density scenarios. This
problem is eliminated in UMBP, where a multidirectional
broadcast is employed at intersections to conduct the
forwarding vehicle selection in different directions. The
evaluation of UMBP is performed in terms of dissemination
delay, reception rate, and propagation speed, while data
redundancy overhead is not evaluated. Another example is
the Road-Casting Protocol,whichassigns differentbroadcast
probabilities to receiving vehicles based on the distanceand
the link quality. Vehicles with higher probabilities have
lesser waiting times before rebroadcasting.
The obtained results have proven the efficiency of
Road-Casting in terms of data delivery ratio and end-to-end
delay. However, the overhead under high densities is still
high. Focus on data redundancy reduction by selecting the
appropriate disseminators. The selection process relies on
complex network metrics, which are: degree distribution,
centrality and clustering, travel time and distance.
Nevertheless, the measurements of these metrics require a
comprehensive network analysis that is computationally
expensive. The authors assume that a prior density
knowledge is available to drivers based on mobility traces
studies, which represents a major drawback of their
proposal. Similarly, data redundancy is reduced in HBEB,
where a beacon-free algorithm isemployed.Thealgorithmis
designed to form multiple backbones of relay vehicles in
VANET. This formation poses an extra dissemination delay,
which makes it inappropriate with safety-related
applications in the context of ITS.
3. GREY WOLF OPTIMIZER FUZZY ALGORITHM
Grey Wolf Optimizer. The Grey Wolf Optimizer
algorithm (GWO) is a meta-heuristic that was originated in
2014 created by Seyedali Mirjalili, and inspired basically
because in the literature there was not a Swarm Intelligence
(SI) technique based on the hierarchy of leadership of the
Grey Wolf.
This presents aspects concerningthetuningoffuzzy
controllers (FCs) by grey wolf optimization (GWO)
algorithms with focus on cost-effective Takagi-Sugeno
proportional-integral fuzzy controllers (T-S PI-FCs). GWO is
one of the latest swarm intelligence algorithms, which has
been developed by mimicking grey wolfsocial hierarchyand
hunting habits. T-S PI-FCs are applied to servo systems,
represented as non-linear processes characterized by
second-orderdynamicswith anintegral component,variable
parameters, a saturation and dead-zone static non-linearity.
The variable parameters of the process justify the need to
design fuzzy control systems with a reduced process
parametric sensitivity. Four optimization problems are
defined with this regard, with the tuning parameters of T-S
PI-FCs considered as vector variables and with objective
functions that include the weighted output sensitivity
function of the state sensitivity model with respect to
process parametric variations. GWO is next employed in the
minimization of these objective functions. Simulation and
experimental results are given for a case study that deals
with the optimal tuning of T-S PI-FCsfortheangularposition
control of a laboratory non-linear servo system.Theprocess
gain is variable, and fuzzy control systems with reduced
process gain sensitivity are offered. Energy is a valuable
resource in Wireless Sensor Networks (WSNs).
The status of energy consumption should be
continuously monitored after network deployment. The

information about energy status can be used to early notify
both sensor nodes and Network Deployers about resource
depletion in some parts of the network. It can alsobeused to
perform energy-efficient routing in WSNs. In this paper, we
propose a neural network based clustering and energy
efficient routing in WSN with theobjectiveofmaximizingthe
network lifetime. In the proposed scheme, the problem is
formulated as linear programming (LP) with specified
constraints. Cluster head selection is done using adaptive
learning in neural networks followed by routing and data
transmission. The simulationresults showthattheproposed
scheme can be used in wide area of applications in WSNs.
A. Assumptions and Requirements
For selecting the clusterheadweuseGWOandANN.
Grey wolf optimizer uses the leadership hierarchy. In the
leadership four types of grey wolves layers are used alpha,
Beta, Delta, Omega. The nodes in the layer 1 are called as
Leader, nodes in the layer 2 are called as Co-leader, the
nodes in the layer 3 are called as Elder and the nodes in the
layer 4 are called as Members. By this approach power
consumed is diminished and the network lifetime gets
increased. If the cluster has only one node in the layer 1
means, then the node will be selected as a cluster head
directly. If the two node are present in the cluster, then the
cluster head (CH) is selected by using Residual energy of
both the node, the node which has high residual energy is
elected as cluster head If the cluster has more than two
nodes, then the cluster head is selected by using Gaming
theory with some parameters such as Residual energy,
Packet reception ratio.
Neural networks are a set of algorithms , modeled
loosely after the human brain , that are designed to
recognize patterns Neural network is used to predict the
cluster head and it increases energy efficiency. They
interrupt sensory data through a kind of machine
perception, labeling or clustering raw input.
In the following we define the basic terms usedinthispaper:
• BROADCAST INITIATOR: is the node which originates a
new data and intends distribute it to nearby node.
• REALY NODE: is the node which updates traffic condition
before rebroadcasting a data that was originally initiated by
another node.
• AREA OF EVENT (AoE): where the data is initiated to
indicate the event.
• AREA OF INTREST (AoI): is the wide area in the selected
area where data should deliver with highest possible ratio.
• APPROACHING NODES: the node moving towards theAoE.
• REEDING NODES: the node moving away from the AoE.
1. Find the Eigen values and Eigen vectors of the nodes
2. Partition the graph as two clusters by using positive and
negative Eigen value.
If (Eigen vector (i) == positive)
Node (i) == Cluster
else
Node (i) == Cluster 2
end
After the first iteration ‘n’ number of clusters were formed.
(n = 2 when k = 1)
3. For further cluster partitioning
Number of nodes inside the cluster (M)
Grouping the nodes and form a cluster with less
energy consumption by that maximizing the life time is an
challenging task in WSNs. The two important steps in
clustering are Cluster formation and Cluster Head (CH)
selection. The novel and efficient clusteringcalledClustering
using Eigen Values (CEV) is proposed in this paper with the
increased lifetime of the sensor nodes using the spectral
graph theory. This work uses the Laplacian matrix of
spectral theory for clustering. The Eigen values of Laplacian
Matrix and its corresponding eigenvector are used to group
the nodes of WSN. CH is selected using fuzzy logic and
constraints on energy and distance. This work is evaluated
and compared with LEACH and HEED for performance
comparison. The results obtained in this work show thatthe
proposed work yields better. The grey wolves follow very
firm social leadership hierarchy. The leaders of the pack are
a male and female, are called alpha (α). The second level of
grey wolves, which are subordinate wolves that help the
leaders, are called beta (β).Deltas (δ) are the third level of
grey wolves which has to submit to alphas and betas, but

dominate the omega. The lowest rank of the grey wolf is
omega (ω), which have to surrender to all the other
governing wolves
This algorithm works by assigning membership to
each data point corresponding to each cluster center on the
basis of distance between the cluster center and the data
point. More the data is near to the cluster center more is its
membership towards the particular cluster center.
Summation of membership of each data point should be
equal to one. After each iteration membership and cluster
centers are updated according to the formula:
Proposed Network
Setup Parameters Value
Network Area 100,100
Base Station (x,y) 50,50 or 50,150
Number of nodes 100
Initial Energy 0.1 joule
Transmitter Energy 50*10-9
Receiver Energy 50*10-9
Free Space (amplifier) 10*10-13
Multipath (amplifier) 0.0013*10-13
Effective Data
Aggregation
50*10-9
Maximum Lifetime 2500
Data packets size 4000
The comparison is done in the termsofdata packets
received at BS. The graph expounds that in proposed work,
the amount of packets received at base station ishigherthan
the number of packets received at base station in ACOPSO.
The alive nodes in proposed work are higher than
the number of alive nodes in traditional work.
Assumptions made for this routing:
(i) Sink (Base Station) is located inside the sensor field
(ii) Unlimited resource is allotted for BS.
(iii) After deployment sensor nodes are unattended. So that
recharging or changing of battery is not possible.
(iv) Links are asymmetric because of the mobility of the
nodes.
(v) All sensor nodes are not equipped with any location
finding device.
(vi)Mobility of the node is controllable and predictable.
It is noticed that the power consumption is less in
GWO approach compared to LEACH, HEED and CHEATS
because of reduction in number of iterations and the
hierarchy followed in proposed approach. Thus, the lifetime
of the entire network gets increased.

It is observed that the network life time of Grey-
Wolf optimization approach is increased by 10% compared
to CHEATS and 12% compared to HEED and 16% compared
to LEACH.
The existence of theconnectivity betweenthenodes
as strong even the transmission range gets increased.
It is observed that data packets delivered effectively
in GWO approach compared with other approaches.
4. CONCLUSION
In wireless sensor network due to proper routing
technique the power consumption can be reduced
reasonably. Due to this, the lifetime of the entire network
gets enhanced and the node which is responsible to forward
the packet plays a main role in routing. In proposed Grey
Wolf Optimization (GWO) approach a proper cluster head is
elected using a layer based architecture that splits up the
entire region into four layers, each having several
responsibilities and the number of iterations is reduced
compared with LEACH, HEED, CHEATS and GTDEA. This
approach guaranteed that a considerable amount of
reduction power consumption, thereby increasing the
lifetime of the entire network. Simulations performed
showed that the energy consumption, throughput and
network lifetime has improved compared to other
approaches.
REFERENCES
[1] Ian F. Akyildiz, Tommaso Melodia, Kaushik R. Chowdury,
“Wireless Multimedia Sensor Networks: A Survey” IEEE
Wireless Communications, December 2007.
[2] Ian F. Akyildiz, Tommaso Melodia, and Kaushik R.
Chowdhury, “Wireless Multimedia Sensor Networks:
Applications andTestbeds”, ProceedingsoftheIEEE,October
2008.
[3] Ian F. Akyildiz, Tommaso Melodia, Kaushik R.
Chowdhury, “A survey on wireless multimedia sensor
networks”, Computer Networks, Elsevier Science BV, 2006.
[4] I. F. Akyildiz, T. Mеlodia and K. R. Chowdhury, “A Survey
on WirelеssMultimеdiaSеnsorNеtworks,”
ComputеrNеtworks, Vol. 51, No. 4,
[5] Islam T.Almalkawi, Manel Guerrero Zapta, Jamal N.Al-
Karaki and Julian Morillo-Pozo,
“WirelеssMultimеdiaSеnsorNеtworks: Current Trends and
Future Directions” Sensors 2010.
[6] Ameer Ahmed AbbasiandMohamedYounis,“Asurveyon
clustering algorithms for wireless sensor networks”,
Computer Communications, Elsevier Science BV, 2007.
[7] Ossama Younis, Marwan Krunz, and Srinivasan
Ramasubramanian, “Node Clustering in Wireless Sensor
Networks: Recent Developments and Deployment
Challenge”,IEEE Network, May/June 2006.
[8] Xuxun Liu, “A Survey on Clustering Routing Protocols in
Wireless Sensor Networks”Sensors 2012, 11113-11153.

IRJET- Efficient Data Dissemination using GWO Fuzzy in Neural Networks

Recommended

More Related Content

What's hot (17)

Similar to IRJET- Efficient Data Dissemination using GWO Fuzzy in Neural Networks (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET- Efficient Data Dissemination using GWO Fuzzy in Neural Networks