Energy Aware Talented Clustering with Compressive Sensing (TCCS) for Wireless Sensor Networks

International Journal of Computer Networks & Communications (IJCNC) Vol.14, No.4, July 2022
DOI: 10.5121/ijcnc.2022.14404 57
ENERGY AWARE TALENTED CLUSTERING
WITH COMPRESSIVE SENSING (TCCS) FOR
WIRELESS SENSOR NETWORKS
Bejjam.Komuraiah1
and Dr. M.S. Anuradha2
1
Research Scholar, Department of ECE,
AU College of Engineering (A), Visakhapatnam, AP, India 530003
2
Professor, Department of ECE,
AU college of Engineering (A), Visakhapatnam, AP, India 530003.
ABSTRACT
Wireless sensor networks (WSNs) are networks of sensor nodes that interact wirelessly to gather
information about the surrounding environment. Nodes are often low-powered and dispersed in an ad hoc,
decentralized manner. Although WSNs have gained in popularity, they still have several serious
shortcomings, like limited battery life and bandwidth. In this paper, the cluster head (CH) selection, the
Compressive Sensing (CS) theory, the Connection-based Decentralized Clustering (CDC), the relay node
selection, and the Multi Objective Genetic Algorithm (MOGA)are all taken into account The initial stage
provided a theoretical revision to the concepts of network construction, compressive sensing, and MOGA,
which impacted the improvement of network lifetime. In the second stage developed a novel model such as
Energy Aware Talented Clustering with Compressive Sensing (TCCS) for the sensor network. This
approach considers increasing longevity but also raises the network's overall quality of service (QoS). In
the analysis, the TCCS model is applied to both the centralized and distributed networks and compared
with the existing methods. When compared to the previous methods, the simulation results show that the
proposed work performs better in terms of the calculation of maximum packet delivery ratio of 93.93
percent, minimum energy consumption of 8.04J, maximum energy efficiency of 91.04 percent, maximum
network throughput of 465.51kbps, minimum packet loss of 282 packets, and minimum delay of 63.82 msec.
KEYWORDS
Wireless sensor networks, clustering, clusterhead, compressive theory, decentralized clusting, relay node.
1. INTRODUCTION
Wireless sensor network exhibits significant advantages in terms of minimal cost, tiny sensor
nodes and small-scale factor. It can be employed in the dangerous and cumbersome area for
region monitoring and controlling, with the automated mundane tasks for deployment. The
conventional sensor units are expensive and fails to perform effective computational and
communication capabilities for the present sensor nodes [1]. The present sensor node involved in
data sensing, processing, storing and forwarding those are powered with the battery. The vast
range of application are implemented with the wireless sensor network with minimal cost
solutions for monitoring environment for different applications such as military, target detection,
application tracking and civilian, precision farming, health care application. Also, WSN has been
applied in residential application for the management of energy, safety and efficiency of the outer
space explorations [2].

58
Over the period of time, Wireless sensor network are deployed over the region with motes which
provides the short-range radio communication and minimal coverage area [3]. The WSN
comprises of large number of nodes those involved in provision of multi-hop communication in
the network and collaboration for each node to provide appropriate coverage and connectivity.
Apart from the traditional concern ad-hoc environment involved in collaboration and
communication with the plagued power and energy management for the battery-operated nodes
[4]. Sensor nodes are non-rechargeable those are equipped with batteries therefore energy
efficiency is considered as important concern for increasing the lifetime of the network. In WSN
environment, energy consumption and modeling are major concern for the design and
implementation of the network energy optimization. The sensor node energy consumption
depends on the three factors such as sensing, signal processing and communication [5]. The
communication module comprises of the signal transmission and reception which consumes more
energy. Hence, majority of the research concentrated on the reduction of energy consumption to
minimization communication cost of the network.
The key factor involved in management of the drained energy is effective management of the
network coordinates between sensor nodes through clustering. To minimize the data transmission
time and energy consumption in the network the small sensor nodes are grouped together into
small groups known as clusters [6-10]. Also, the grouping of sensor nodes iscalled as clustering.
Within every cluster lead node is elected act as a cluster head (CH). The CH in the sensor node
exhibits higher capabilities rather than the other sensor in the network. Within the respective
cluster the cluster head is selected for the sensor node where CHs are pre-assigned by the user.
The clustering exhibits the advantages involved in transmission of the aggregated data in sink or
base station. The process of clustering provides scalability for large nodes and reduce energy
consumption. In the centralized clustering process the cluster head is fixed. Through this with
failure of the cluster head the complete cluster is collapsed. Hence, it defines centralized
clustering does not have adequate reliability. However, distributed clustering involved in
provision of the data reliability when cluster head fails. In this case, the complete network fails
when all nodes are failed. In WSN environment distributed clustering commonly engaged in
provision of efficient gathering of data and reliability [25-27].
The theory of Compressive Sensing (CS) offers an efficient paradigm for WSN data gathering.
The signal may be captured at a sample rate lower than the Nyquist sampling rate using CS. With
the decrease in sample frequency, the CS technique may be used in the network to increase the
network's lifespan. There are minimum linear measurement features required to reconstruct the
sparse signal using the CS-based techniques. The l1-norm convex optimization problem may be
used to locate the signal source. CS theory is concerned with determining sample frequencies
based on the features of the signal rather than the bandwidth of the signal. One of the most
difficult issues in WSN design and deployment is reducing the network's WSN energy usage. The
components of the network's energy consumption must be evaluated in order to solve the issues
associated with WSN's energy consumption. Additionally, the development of a viable model for
analyzing WSN energy usage has been aided by many suggested models, all of which are linked
to the CS method.
Contributions:
 It is suggested to do energy-aware talent grouping using Compressive Sensing (CS)
approaches. The developed model is regarded as a useful tool for the analysis of energy
usage using the CS-based data collection approach to improve network efficiency.
 The proposed method considers the Cluster Head selection, Compressive Sensing (CS)
theory, Connection-based Decentralized Clustering (CDC), Relay node selection, and
Multi-Objective Genetic Algorithm (MOGA) for energy-efficient path selection.

59
 Cluster selection (CH) uses Euclidean distance to decrease network energy use.
Connection-based Decentralized Clustering considers data transport hops and message
weight..
 Through relay node selection the consumption of energy is reduced. For energy
efficiency and reconstruction error, Multi Objective Genetic Algorithm (MOGA) is
utilized.
 The suggested strategy enhances the network's overall Quality of Service (QoS), in
addition to considering lifetime enhancement.
The present paper is organized as follows: In section 2 presented a related works for energy
aware clustering along with compressive sensing technique. The network model considered for
analysis is presented in section 3. The proposed CH selection and clustering in the WSN model is
elaborated in Section 4. The proposed model involved in formulation of the relay node selection
and multi-objective genetic algorithm. The performance of the proposed model is presented along
with comparison of TCCS in section 5. The overall conclusion for the proposed model is
presented in Section 6.
2. LITERATURE REVIEW
The wireless sensor node's lifetime is inversely proportional to its battery consumption. By
calculating the energy consumption of the network's nodes, applications and routing protocols
may make better judgments to increase the sensor network's lifespan. This section provides the
existing algorithmutilized for reduced energy consumption and reliable routing scheme for the
WSN. KalpnaGuleria et al.,[11] proposed an scheme for minimization of the energy consumption
through Enhanced Energy Proficient Clustering (EEPC) for complete sensor for the application
of field tracking. The constructed nodes are designed based on the mobile and fixed nodes.
Initially, the nodes those are fixed involved in information broadcast and select mobile nodes
form the cluster head with the fixed nodes. The nodes those are mobile elects the cluster head
(CH) with the associated energy and placement level of nodes. With Mobile sensor nodes (SNs)
data has been transmitted to the CH with the introduction of concept relay nodes, with fixed
nodes. The developed EEPC algorithm selects the relay node based on the calculation of the
fitness value based on location and velocity. The developed scheme involved in minimization of
energy depletion to improve network lifetime. However, the proposed methods fail to evaluate
the transmission delay for data transmission.
V. Seedha Devi et al., [12] proposed a Cluster-based data aggregation reduces latency and packet
loss in WSN. Construction of Aggregation Tree and Slot Scheduling are aspects of the proposed
strategy. Every cluster head compresses node data at first. Through the built aggregate tree sink
node employs MST (MST). Based on allotted timeslots and priority, phase 2 aggregated data may
minimize packet loss and latency in the network. The presented approach reduces WSN
retransmission and waiting times, improving network performance. The simulations showed that
the suggested technique reduces delay and overhead while increasing packet delivery rate and
residual energy. However, the planned plan doesn't assess energy use.
Pakdaman Tirani, Shima & Avokh, Avid [13] constructed a Energy-aware CS-based Data
aggregation and "Energy-balanced High level Data aggregation Tree" The technique improves
network longevity and energy balance by considering diverse nodes. The suggested technique is
compatible with mobile sinks and improves network longevity. The developed EHDT technique
featured network load-balancing with weighted routing tree for data transmission and cluster
head creation with sink via effective energy distribution between nodes. Numerical examination
of the suggested method showed that it boosted network transmissions, energy consumption, and

60
longevity. The designed approach has simple sink mobility. The suggested strategy doesn't
optimize sink node mobility.
Senouci, Mustapha & Mellouk, Abdelhamid [14] evaluated a uncertainty aware cluster-based
deployment approach for the sensor environment. The developed scheme comprises of the
different factors based on consideration of the different characteristics for real-world applications
in sensor measurement, sensor spatial measurements, harsh deployment model, reliability and
unreliable connectivity. The developed results expressed that proposed scheme exhibits
significant performance for the real-world wireless sensor network to achieve desire network
performance.
Osama MohdAlia [15] developed an energy-efficient network model for dynamic reallocation of
the mobile BS within cluster-based infrastructure network with the harmony search algorithm. At
first, the developed model allocated information between nodes for framed optimal cluster
number with formulation of the appropriate cluster. Through optimal CHs formulation in the
sensor cluster order are evenly distributed among sensor with formation of CH. The infrastructure
of dynamic environment is evaluated based on the consideration of the number of alive nodes and
load balancing in the sensor node. Subsequently, with the optimal placement of the BS CH is
determined to minimize the communication distance in the network. Finally, the data
transmission and sensing placed based on placement of each sensor node with CH and aggregate
the data sensed in BS. The simulation results expressed that proposed scheme exhibits improved
lifetime of network, data delivery and energy consumption for static and random BS in the
network models.
Anis Jari and Avid Avokh [16] constructed a plan for clustering, multilink placement, and load
balancing. The suggested technique includes Multi-sink Placement and Anycast Routing (MPAR)
and EMPAR. The proposed MPAR and EMPAR architecture has clustered sensors. Each sensor
node communicates data to the cluster head (CH) through load-balancing tree. In upper level, the
approach uses a modified particle swarm optimization technique to find the ideal sink node
placement. The ant colony algorithm optimises the routing tree based on a high-level analysis.
Every tree for anycast employs compressive sensing (CS) for data forwarding and aggregation.
Simulations show that the suggested system improves efficiency, network longevity, energy
consumption, and variation.
Lv, Cuicui& Wang, Qiang& Li, Jia. [17] constructed a framework to minimize the energy
consumption in WSN. The constructed framework uses the covariance matric for sparsify the
generated sensor data. The developed scheme uses the numerical sparsity to evaluate the
performance of data. The constructed matrix in the network comprises of the sparse binary
measurement with computation of numerical sparsity. For every measurement, the part of sensor
nodes is involved in gathering of sensor data and transmit data for sink node to recover data. The
experimental analysis demonstrated that the constructed sparsify exhibits real temperature data
approximation. Compared with other types of constructed sparsifyingexhibits numerical
temperature data those are smaller and sequential temperature performance recovery of the data.
The proposed framework model exhibits minimal total energy consumption of the proposed
scheme is minimal than the other compressive and data gathering algorithm. The proposed
scheme exhibits limitation of the higher time complexity for the reduced energy consumption.
Zhang, Ce at al.,[18] constructed a algorithm for data gathering to minimize energy consumption
and packet loss. The proposed model involved in formulation of the cluster head with Sparest
random measurement matrix (SRMM) through the received data to minimize the lost node
measurement and reduce the measurements. To employ between cluster spatial correlation is
performed for the constructed sink with the block diagonal matrix (BDM) through reconstruction

61
of the SRMMs for the entire network data. Additionally, with the formulation of the optimal
cluster number the developed model reduces the power consumption. The proposed SR-BDM
involved in estimation of the emulated data for the sensor data GreenOrbs respectively. The
simulation results demonstrated that the proposed model exhibits higher precision through
reliable links and packet link of 60% with minimal energy consumption.
G. Yang, M. A et al.,[19] developed a model for energy efficient communication for green
communication with the hierarchical approach based on clustering to monitor health status of
patients. The developed model organizes the devices in the cluster with devices in equal sizes.
Within every cluster the designated cluster collect data from the member devices to broadcast
information with centralized base station. The constructed model evaluates the energy
consumption of the devices at different states such as idle, sleep, awake and active to perform
data transmission between two different states. Through analytical modeling energy consumption
of each device at different states are measured. The simulation results demonstrated that the
analytical approach improves the lifetime of the network with reduced energy consumption at
different states.
S. M. M. H. Daneshvar et al.,[20] constructed a new effective clustering algorithm for election of
cluster head with grey wolf optimizer (GWO). The GWO is class of intelligence algorithm based
on the consideration of the grey wolf’s behavior with comparative examination of results. The
proposed model utilizes the similar clustering with proposed protocol with the formulated
consecutive rounds. The designed protocol concentrated on the energy saving those are utilized
for the clustering reformation. The dual-hop routing strategy is used in the provided model to
choose the CH depending on the base station's distance. The analysis of the results expressed that
the proposed scheme exhibits reduced and balanced energy consumption through single-hop
communication. The performance of the designed protocol expressed that the proposed scheme
increases the lifetime of the network with the similar protocols.
M. A. Mazaideh and J. Levendovszky [21] constructed a model for WSN to perform efficient
data transmission data nodes with the CS-based approach. The developed model uses the energy
efficient optimization model with the multiple objective genetic algorithms (MOGA) to optimize
the transmission range measurements with the matrix sensing scheme. The developed model
aimed to strike the balance between accuracy and energy efficiency. Through the optimized
values the constructed paths are evaluated based on muti hop manner. The numerical analysis and
experiments stated that expressed output model exhibits MOGA to elect the appropriate
combination based on the measurement number and fitting in the transmission range. The
analysis stated that the developed model exhibits the effective balance between the accuracy and
energy efficiency. The experimental results demonstrated that existence of the measurement
matrices exhibits minimal lower coherency and improved accuracy of CS. Ghaderi, M.R et
al.,[22] presented a complete model for analysis of the CS-based energy consumption in the
WSN. The designed model expressed that source energy consumption is based on the CS those
can be divided in to two categories such as communication and computation energy those are
modeled based on the energy components. The distinct number of the data aggregation schemes
are developed and presented based on CS-based in WSN are investigated and discussed
comparatively. The developed model expressed that the proposed optimization scheme perform
effectively in the designing of the CS-based operation in WSN.
Q. Wang, D et al.,[23] proposed a compressive sensing-based (CS-based) for the clustering
strategy. The relationship between the two WSN adjacent layer are exhibited as lemmas, cluster
optimal size, cluster head (CH) optimal distribution and corresponding analysis are evaluated.
Additionally, the problem of the hot spot is designed with the minimization of the energy
consumption resulted in the rotation of the CH role. Finally, the backup CH (BCH) exhibits the

62
mechanism to compute the roles and functionality of the CH and BCH. Subsequently, the
performance of the network is developed with the energy-efficient compressive sensing-based
clustering routing (EECSR). With extensive simulation experiments energy performance are
evaluated in the different aspects. Through extensive simulation analysis the evaluation and
experiments are examined for energy performance. The comparative analysis is performed with
the CS-based and clustering algorithm to evaluate the impact of the EECSR to increase the
energy efficiency and improves the lifetime of the WSN.
In this paper the proposed scheme concentrated on the alleviating the challenges associated with
the issues involved in energy consumption. Primarily, the developed theoretical model involved
in evaluation of the concepts related to network construction, MOGA and compressive sensing
those impacts on the network lifetime. Secondly, the proposed Energy Aware Talented Clustering
scheme with the Compressive Sensing (TCCS) in WSN is evaluated. The proposed TCCS
comprises of the optimal selection relay nodes to minimize the CH node energy depletion. The
proposed algorithm effectively minimizes the data transmission complexity and improved
network lifetime.
3. PROPOSED METHOD
Battery-powered sensor nodes have a short life expectancy and are thus very energy-reliant.
Research into reducing the power consumption and the overall size of wireless sensor networks
(WSNs) is critical if these networks are to be widely deployed.
Figure 1. Architecture of the proposed method
The proposed method is sub divided into four main sections namely Network basic models, Multi
objective Generic Algorithm (MOGA), Clustering Module with decentralized Structure and
Relay Node Selection. Multi-objective GA consist of energy efficiency calculation and
reconstruction error parameters. Finally, MOGA based Compressive Sensing with Performance
analysis. The network basic model consists of network model, energy consumption model and

63
compressive sensing. With Genetic Algorithm Multi-Objective concept is combined in MOGA.
Clustering involves the construction of clusters, choosing cluster leaders, and using a
decentralized method with relay node selection. The suggested technique architecture is shown in
figure 1.
The Euclidean distance is calculated in this case to pick the Cluster Head, which significantly
lowers the network's energy usage. The connection based decentralized clustering model give
attention to the number of hops used for data transfer and the weight of the message. through
relay node selection the consumption of energy is reduced. Finally, MOGA with compressive
sensing consists of energy efficiency and reconstruction error. Measurements of metrics
including the packet delivery ratio, energy efficiency, energy consumption, end-to-end latency,
network throughput, and packet loss are used in the calculations for performance analysis. The
suggested work's flow diagram is shown in Figure 2.
Figure 2. Workflow of the proposed TCCS methodology
The detailed explanation of the proposed work is given below.
3.1. Network Model
Consider the battery-powered nodes as N in the WSN those are deployed randomly over the
deployment area size 𝑆 = 𝑎 × 𝑎. The network topology for cluster is employed in the network
with partition of the h clusters. The orthonormal basis comprises of the FFT adopted with sparse
representation and reconstruction algorithm model with the orthogonal matching pursuit (OMP)
method. Consider the basis sparse representation as 𝚿 = (𝜙𝑖,𝑗)𝑁×𝑁
and measurement matric is
denoted as 𝚽 = (𝝋𝑖,𝑗)𝑀×𝑁
. The node in the network is considered as dead when energy level are
completely exhausted.
3.2. Energy Expenditure Model
In the proposed model, the energy consumption in radio with first order is defined as in equation
(1) and (2)
ETx(L, d) = Eelec × L + 𝜖amp × L × d2
, (1)
ERx(L) = Eelec × L, (2)

64
In above equation (1) and (2) transmitting energy consumption for the L-message for the distance
d is represented as ETx(L, d). The energy consumption for the L-bit message is denoted as ERx(L).
The transceiver required energy for transmission of electrical energy is denoted as Eelecand
transmission amplifier for the energy consumption is denoted as 𝜖amp.
3.3.Compressive Sensing Background
In the compressive sensing the reading of N sensor is represented as 𝑿 = (𝑥1, ⋯ , 𝑥𝑁)T
and the k -
sparse is denoted as 𝚿 ∈ ℝ𝑁×𝑁
: represented as in equation (3)
𝑿 = 𝚿𝜽, (3)
where 𝜽 ∈ ℝ𝑁
denoted the sparse basis of the vector coefficient in the sparse basis 𝚿. The k-
sparse X is denoted as the compressive in the vector is denoted as 𝜽with nonzero components as
𝑘(𝑘 ⩽ 𝑁) and the smallest components are defined as (N- k) those can be ignored.
Assume the 𝜽 ∈ ℝ𝑀𝑥𝑁
measurement matrix those are uncorrelated with the basis 𝚿. Then,the CS
measurement variable with X can be denoted as in equation (4)
𝒀 = 𝚽𝑿 = 𝚽𝚿𝜽 = 𝚯𝜽, (4)
Where, sensing matrix is denoted as M << N, and Θ = ΦΨ. The reconstructed original signal X
can have probability of overwhelming for the measurements M for the l1-norm minimization
denoted as in equation (5)
𝑿
̂ = arg 𝑚𝑖𝑛 ∥ 𝑿 ∥1 subject to : 𝒀 = 𝚽𝑿 (5)
Where, the reconstructed sparse signal X is represented as 𝑿
̂
The two factors considered for reconstruction of X from Y that need to be considered: 1) X is
compressive at Ψ, and 2) 𝚽need to evaluate the RIP denoted as 𝑀 ⩾ 𝑐𝑘lg(𝑁/𝑘). In other words,
the condition for k-sparse X is expressed as conditions in equation (6)
(1 − 𝜀) ∥ 𝜽 ∥2
2
⩽∥ 𝚽𝜽 ∥2
2
⩽ (1 + 𝜀) ∥ 𝜽 ∥2
2
(6)
Where 𝑐, 𝜀 ∈ (0,1) while 𝚽 sto withstand the RIP parameter 𝜀.
3.4. Multi Objective Genetic Algorithm
To perform optimization of the different parameters multiple objective genetic algorithms
(MOGAs). The metaheuristics algorithm uses the MOGA those are applied in the WSN to
achieve optimal contradictory objectives. The set of restriction are adjusted based on the
simultaneous objectives. Those objectives are adjusted based on the simultaneous restriction’s
subjects. With multi-objective optimization for optimal solution specific definition are presented
instead of yielding single solution through conventional genetic algorithm (GA). The set of
MOGA non-dominated solutions are accepted to derive the best solution with subjective those
need to be formulated.
MOGA involved in formulation of vector fitness function denoted as F(x) = [f1(x), f2(x), · · ·,
fn(x)]T to derive the decision variable to compute the inequality and equality constraints, through
those constraints the viable domain are defined to derive acceptable solutions.

65
The pareto-Optimality in the MOGA involved in computation of the vector fitness function to
derive the Parento-Optimal solution. The vector in the multi-objective optimization with Pareto-
Optimal solution vector is defined as X = [x1, x2, · · ·, xn] T
those engaged in provision of the
feasible solutions to minimize the least one objective to increases the characteristics of other
objectives. The vector solution of Pareto-optimal set or Pareto-front 𝐗
̂ = [𝑥
̂1,𝑥
̂2,⋯ , 𝑥
̂𝑛]𝑇
those
are not dominated for any other solution space where Xˆ are dominated for the value X if and
only if as presented in equation (7) and (8)
∀𝑖 ∈ 𝑖 = 1,2, ⋯ , 𝑛, 𝑓𝑖(𝑥) ≤ 𝑓𝑖(𝑥
̂) (7)
And
∃𝑖 ∈ 𝑖 = 1,2, ⋯ , 𝑛, 𝑓𝑖(𝑥) < 𝑓𝑖(𝑥
̂) (8)
4. ENERGY AWARE TALENTED CLUSTERING APPROACH USING
COMPRESSIVE SENSING (TCCS)
In energy saving scheme the transmission count is reduced but it is not sufficient enough.
According to first-order radio model for the energy consumption for transmission of the L-bit
messages based on the computation of the distance leis between the nodes. The figure 3 presented
that the comparative examination of the aggregation trees with the 7 nodes in which the near link
is computed based on the link Euclidean length. In figure Figs. 3-a and 3-b presented the total
transmitted packets involved in transmission. In example of both case node 0 involved in
transmission of packets 5 to the sink node. However, the computation of the links between Fig. 3-
b is less than the Fig. 3-a for computation of the energy consumption. The main objective of the
network to maximize the lifetime of the network, in this paper both transmission and remaining
network energy are considered. At the end, the proposed scheme involved in resolving cluster-
based routing problem, CS theory, sink placement and load balancing [24-26].The data
aggregation is constructed based on the trees with round by round.
Figure 3. Process for aggregating data using TCCS
A cycle that involves calculating the amount of time needed to transport the aggregated data to
the sink and gather data using sensor nodes. Each round in the developed model comprises of the
three phases: At first phase, the sensor node effectively forms the clusters. In second phase,
remaining energy is determined based on the CHs energy and position based on the previous
round. In third phase, every node transmits the data with the corresponding CH through
formulation of the shortest path tree. With the CS scheme the data collected is compressed.
Further, the projection of every CH generates the forwarded information towards sink by sink
with the routing tree in high-level scenario. Based on the following constraints different phases
are proposed for TCCS those are explained as follows:
4.1. Clustering Model
As stated above, the sensor nodes are partitioned into C clusters with the implementation of the
significant strategy modifications. The proposed scheme uses the random set of the CHs election.

66
The nodes those are near to the CH are connected and formulate the cluster. In each cluster, the
nodes those are near to the CH are elected [27-29]. In every cluster, a CH selects in such a
manner to summarize the hop counts in the cluster nodes to minimize the CH. The similar
process is continued till no changes are observed in the CHs. At first round, the random CH is
stated compatibility with the derived energy consumption model based on the Euclidean metric
distance with hop count. The better CHs involved in minimization of the energy consumption.
The objective to perform effective network load-balancing with consideration of the CH. With
the typical node Euclidean distance calculation typical node are minimal than the CH distance,
the sink cluster member are computed for the node values. This involved in minimization of the
transmission number and improves the load balancing for the sink node.
Thus, the steps involved in processing are presented as follows:
1) The distance of Euclidean between CH and other nodes are computed. Every sensor nodes
become the cluster member based on the Euclidean distance estimation for the minimized CH.
2) In each cluster, the selected CH is based on the summation of the Euclidean distance between
nodes and CH in minimized manner.
3) The above process continues until the no changes are observed in CHs.
4.2. CH Selection
The constructed traffic model of the sensor node computes the data measured based on round by
round. The CHs involved in utilization of the more energy than the other sensors, these impacts
on the lifetime of the network. Thus, the proposed model involved in computation of the different
node energy distribution. This leads to uniform loss of energy with increased First Node Dies
(FND). In first round, the CHs determined the strategy that explained in phase 1. In next round,
based on the first metrices, the hop between 1-hop or 2-hop distance are involved in computation
of the present CH, the candidate node computes the more energy required for the CH as
explained in equation (9)
CH1(𝑗) = arg 𝑚𝑎𝑥
𝑖∈cos didate nodes (𝑗)
Er(𝑖) (9)
Where, in above equation (9)CH1(𝑗)denoted the jth cluster CH based on the considered first
metrics and Er(𝑖) illustrate dthe remaining energy in the ith node those need to be randomly
broken. Based on this, the factor considered are remaining energy and distance are presented in
equation (10)
CH2(𝑗) = arg 𝑚𝑎𝑥
𝑖∈𝐶 tuster nodes(j)
(
Er(𝑖)
𝑑𝑚𝑖𝑛(𝑖,𝐶𝐻𝐹𝑅(𝑗))
) (10)
Where CH2(𝑗), 𝐶𝐻𝐹𝑅(𝑗),and 𝑑𝑚𝑖𝑛(𝑖, 𝐶𝐻𝐹𝑅(𝑗))demonstrate the jth cluster head in the jth
cluster of the second metric, jth cluster of CH is computed in first round and minimal hop count
is computed for nodes I and 𝐶𝐻𝐹𝑅(𝑗), respectively.
Based on the second metric, the nodes those are closer to center are having more chance to select
as CH. The reduced energy consumption and transmission number is based on the remaining
node energy, different node energy consumption.
Energy, distance, and delay are factors that are taken into consideration while choosing a cluster
head.

67
The energy consumption between the two nodes is given in below equation (11)
𝑓energy
=
𝑓energy (𝑞)
𝑓energy (𝑝)
(11)
𝑓energy
(𝑞) = ∑𝑗=1
𝑀
𝑢𝐸𝑁(𝑗)
𝑢EN(𝑗) = ∑
𝐿
𝑖=1
𝑖∈𝑗
(1 − EN(𝐷𝑖) ∗ EN(𝐶𝐻𝑗)) ; 1 ⩽ 𝑗 < 𝑀
𝑓energy
(𝑝) = 𝑀 ∗ Max𝑖=1
𝐿
(𝐸𝑁(𝐷𝑖)) ∗ Max𝑗=1
𝑀
(𝐸𝑁(𝐶𝐻𝑗))
where 𝐸𝑁(𝐷𝑖) and 𝐸𝑁(CH𝑗)implies the energy of 𝑖th
the normal node as well as the energy of
𝑗th
the CH, respectively. 𝑓energy (𝑞), refers to the energy between the CH and the normal node
and between the CH and the BS of the network.
Distance is indicated in mathematical Eq. (12). The value of 𝑓dist
(𝑞)should fall within the
category of[0,1].
𝑓dist
=
𝑓dist (𝑞)
𝑓𝑑𝑖𝑠𝑡(𝑝)
𝑓𝑑𝑖𝑠𝑡
(𝑞) = ∑
𝐿
𝑖=1 ∑
𝑀
𝑗=1 ∥
∥𝐷𝑖 − 𝐶𝐻𝑗∥
∥ + ∥
∥𝐶𝐻𝑗 − 𝐵𝑠∥
∥
(12)
where 𝑓dist
(𝑞) shows the distance between the CH and the network's BS, as well as the distance
between the CH and the normal node.
In Eq. (13), the nodes' data transmission delay is described, and the delay value is required to fall
within [0, 1]. As the number of nodes in a cluster decreases, the latency also becomes
significantly decreased.
𝑓delay =
Max(CH𝑗)
𝐿=1
(13)
4.3. Link based Decentralized Clustering (LDC) Approach
In this section, the algorithm for the LDC scheme is developed and pseudocode is presented for
analysis. The developed algorithm is stated with the initiating messages based on the originator’s
node using clustering. The initiated nodes are denoted as O = {O1,O2, . . ., OQ}. The started node
involved in message circulation process with transmission of message to all nearby nodes. Every
message cluster of the tuple comprises of the five fields those are presented as follows:
 Originator ID (OID): The originator node field are provided with the unique identity.
 Message ID (MID): The constructed field classifies all messages form the all messages
through the originators.
 Message Weight (MWeight): The message carried weights for data transmission
 Source ID (SourceID): The message visited by the recent node are indicated in this field.
 Time to Live (TTL): The maximal hop count those need to be recirculated based on the
number of hops.

68
The constructed fields SourceID, the MID, and the OID self-explanatory in the tuple. The
weighted function is estimated based on the probability to reach the destination node from the
source node. The initialized originator node Ol message weights are denoted as 𝑠𝑔. 𝑀𝑊𝑒𝑖𝑔ℎ𝑡 =
1
𝐷𝑒𝑔𝑟𝑒𝑒 (𝑂𝑙)
.
The field of TTL comprises of the small value integer those only constraints the originator node
those uses similar TTL value. Every Vi node compute the value set denoted as
TotalWeight(Vi,Ol). The defined values stated that the all-message weights are generated from
the Oland reaches the destination node as Vi. Upon reception of the message Msg the recipient
updates the Vi as TotalWeight function for the originator message. Every Vi evaluate the TTl
message those are higher than 0. If Viperform data forwarding of messages between nodes. The
recirculation is evaluated based on the recipient MWeight update in the TTL. The weight
messages are classified based on the TTL degree as Vi those are decremented by the value 1. The
message circulation halts are denoted as Vi in TTL as 0 or MWeight those are significantly
minimal. Every node in the network receives multiple messages from the different data nodes.
The computation performs the function of TotalWeight in the originators with the received
messages. The nodes those receives the last message need to wait for few times to ensure high
messages for processing. Then the node involved in the formulation of the originator for the
value of maximal TotalWeight. To construct a particular cluster the messages are originator with
the informing decision based on the node-ID. In case if all nodes lie below TotalWeight values
than the node remains the outlier. The node those are joined in the cluster are decided and remain
in the discover outlier at any point instances. It involved in accumulated weights form the
different originator messages. In those cases, the nodes already in cluster are within the group the
originator cluster make decision about the cluster. It involved in transmission of new messages in
the originator to notify the decision about the cluster.
Algorithm 1: Algorithm Executed by Message Originator 𝑶𝒍
Create a New Message 𝑀𝑠𝑔
𝑀𝑠𝑔. 𝑂𝐼𝐷 ← 𝑂𝑙, 𝑀𝑠𝑔. 𝑀𝑊𝑒𝑖𝑔ℎ𝑡 ←
1
Degree (𝑂𝑙)
𝑀𝑠𝑔. 𝑆𝑜𝑢𝑟𝑐𝑒𝐼𝐷 ← 𝑂𝑙, 𝑀𝑠𝑔. 𝑇𝑇𝐿 ← InitialTT 𝐿
𝑀𝑠𝑔. 𝑀𝐼𝐷 ← Current System Time { A unique value }
for Each node 𝑉𝑖 ∈ 𝑁𝑏𝑟(𝑂𝑙) do
Send 𝑀𝑠𝑔 to 𝑉𝑖
end for
Algorithm 2: Algorithm Executed by Node 𝑽𝒊 on Receiving 𝑴𝒔𝒈
{Check whether I have received messages from 𝑀𝑠𝑔. 𝑂𝐼𝐷}
if I have seen messages from 𝑀𝑠𝑔. 𝑂𝐼𝐷 before then
{Check if the 𝐿𝑎𝑠𝑡𝑀𝑠𝑔𝐼𝑑(𝑂𝑙) == 𝑀𝑠𝑔 ⋅ 𝑀𝐼𝐷}
if LastMsgID(𝑂𝑙) == 𝑀𝑠𝑔 ⋅ 𝑀𝐼𝐷 then
TotalWeight(𝑉𝑖,𝑂𝑙) ← TotalWeight (𝑉𝑖, 𝑂𝑙) + 𝑀𝑠𝑔 ⋅ 𝑀 Weight
else
TotalWeight(𝑉𝑖,𝑂𝑙) ← 𝑀𝑠𝑔. 𝑀𝑊 eight
LastMsgID (𝑂𝑖) ← Msg.MID
end if
else
{This is the first message from 𝑂𝑙}
TotalWeight(𝑉𝑖,𝑂𝑙) ← TotalWeight (𝑉𝑖, 𝑂𝑙) + 𝑀𝑠𝑔 𝑀 Weight
LastMsgID (𝑂𝑙) ← 𝑀𝑠𝑔 ⋅ 𝑀𝐼𝐷
end if
ifTotalWeight(𝑉𝑖,𝑂𝑙) > MaxWeight then

69
MaxWeight ← TotalWeight (𝑉
𝑗,𝑂𝑙)
MaxWeightID ← Msg.OID
end if
if 𝑀𝑠𝑔. 𝑇𝑇𝐿 > 0 and
𝑀𝑠𝑔.𝑀𝑊𝑒𝑖𝑔ℎ𝑡
Degree (𝑉𝑖)
> 𝑀𝑖𝑛𝑊𝑒𝑖𝑔ℎ𝑡 then
Create a New Message New Msg
NewMsg.OID← Msg.OID, NewMsg.SourceI 𝐷 ← 𝑉𝑖
NewMsg.MWeight ←
Msg.MWeight
Degree(V 𝑖)
NewMsg.TTL ← ( Msg.TTL −1), NewMsg.MID ← 𝑀𝑠𝑔. 𝑀𝐼𝐷
for Each node 𝑉
𝑗 ∈ 𝑁𝑏𝑟(𝑉𝑖) do
Send 𝑀𝑠𝑔 to 𝑉
𝑗
end for
end if
Wait for WaitTime in anticipation of other messages
if MaxWeight>WeightThresholdthen
Join the cluster led by MaxWeightID
else
Remain an outlier
end if
4.4. Relay Node Selection
With the selection of CHs, to prevent higher energy depletion CHs are lies far away from the BS,
the developed protocol elects the relay nodes for every node in such as wat the relay is utilized by
the CH. In case if multiple CH uses the relay, then the assigned relay has different timeslots to
handle communication between nodes with the decreased network throughput. Therefore, in the
proposed model relay node uses the only one CH. This leads to lack of relay in CH and directly
transmits the data to the BS. The relay node selection is based on the node distance from the BS
with assignment of appropriate relay node and it proceeds the same procedure for every CHs. To
select the appropriate node relay for CH two characteristics are considered i) reduction of the
toral energy consumption and ii) energy consumption balance between CH and the relay. In
figure 4 presented the hypothetical relay between CH and BS. The distance between BS and CH
are partitioned as equal distance r0 with estimation of relay and BS distance.
Figure 4. Location of CH, Relay node and the Base Station
The proposed model comprises of the dual hop routing and the relation between one-to-one hop
lies between CHs to the relays. The fixed values of the r0with balance and reduction of energy
consumption. In other words, the line segments between point lies are perfect for relay for CH
and BS. To ensure reduced energy consumption data packets need to be delivered between BS
and the CH in which each relay shares equal amount of energy. The proposed model calculates
the r0=1.8169 to select the relay and CH. The relay node between perfect point is lies between
CH and the BS are calculated with the fixed distance ratio r0=1.8169. The point of the CH is the
closest those are not already selected relay. If relay does not have any relay closer to the
threshold value Tr for the perfect time, the CH those does not have relay transmits the BS directly.

70
The developed procedure exhibits different advantages those are presented below:
1) It completely eliminates the hot-spot problem for relay consumes at equal energy amount
of CH where it serves and select the different relays in the rounds.
2) It reduces the amount of energy consumed for packet delivered to BS.
3) It involved in assignment of relays to the different CHs relay as much as possible.
4.5. Multi Objective Genetic Algorithm for Compressive Sensing:
The proposed multi-objective optimization model uses the genetic algorithm to reduce energy
consumption with the error reconstruction. This section presented about the computation of the
reconstruction error and energy efficiency.
4.5.1. Energy Efficiency:
In a WSN network, the transmission distance, represented as d, is used to compute the route loss
exponent α, which is directly proportional to the dα. Statistically, the node transmission range is
defined as R, the communication mean square distance are computed as E[r2
] = R2
/2 [24]. In the
every path in M the average path length is defined as nc, the energy consumption of the node path
is defined as in equation (14)
𝐸𝑐ℎ = 𝑀 ∗ 𝑛𝑐 ∗ (𝑅2
/2)𝛼
(14)
The average distance between path leaders M and BS is represented as dav for the cluster head
total energy consumption for the transmit unit of compressed data denoted in equation (15)
𝐸𝐿𝐵𝑆 = ∑
𝑀
𝑖=1 𝑑𝑎𝑣
𝛼
(15)
The densing field edge square for the D and BS of the centre involved in computation of the
average distance lies between CH and BS as in equation (16)
𝑑𝑎𝑣 = ∫
𝐷
0 ∫
𝐷
0
[(𝑥 −
𝐷
2
)
2
+ (𝑦 −
𝐷
2
)
2
]𝑓(𝑥, 𝑦)𝑑𝑥𝑑𝑦 (16)
where f(x, y) denoted the pdf function joint probability function (pdf), those value is equal to
1/D2 . The total consumed energy between path in the node and CH is presented in equation (17)
𝐸 = 𝑀 (𝑛𝑐 (
𝑅2
2
)
𝛼/2
+ (
𝐷2
6
)
𝛼/2
) (17)
Through the equation (14) it is observed that the increase sin M increases the total energy
consumed.
4.5.2. Reconstruction Error:
The paradigm for the CS reconstruction error is based on the error measured for the original
densed data (eθ), the error in sparse is represented as ex and observed error is denoted as (ey). The
evaluation is based on the consideration of the equation (18) – (20)
𝑒𝜃 =
1
𝑁
∥ 𝜃 − 𝜃
̂ ∥2
2
(18)

71
𝑒𝑥 =
1
𝑁
∥ 𝑥 − 𝑥
̂ ∥2
2
, (19)
𝑒𝑦 =
1
𝑀
∥ 𝑦 − 𝑦
̂ ∥2
2
, (20)
The sensed vector is represented as θ, x, and 4 and the reconstructed vector is represented as
𝜃
̂, 𝑥
̂, and 𝑦
̂. In this paper the (ey) is reduced with the 𝑦
̂ = Φ𝑥
̂, using 𝑥
̂ = Ψ𝜃
̂,if Ω = ΦΨ ,then (20)
becomes:
𝑒𝑦 =
1
𝑀
∥ 𝑦 − Ω𝑥
̂ ∥2
2
(21)
Through equation (21) illustrated that increase in M minimizes the 𝑒𝑦. Table 1 shows the
different variable indications used in the different equations.
Table 1. Different variables indications
Variable Indication
L Messages
d Distance
ETx(L, d). Transmitting energy consumption
ERx(L). Receiving energy consumption
Eelec Electrical energy
𝜖amp Amplifier for the energy consumption
N Sensor
Ψ Sparse basis
R Transmission range
nc Average path length
ex Error in sparse
ey Observed error
eθ densed data
5. SIMULATION RESULTS
The performance ofthe proposed model is evaluated in NS-2 simulation software. End-to-end
delay, packet loss, packet delivery ratio, throughput, energy consumption, and energy efficiency
are the parameters taken in to consideration for the analysis. The performance of the proposed
Energy Aware Talented Clustering with Compressive Sensing (TCCS) algorithm model is
evaluated comparatively evaluated with the existing model such as Compressive sensing-based
energy consumption model (CDAS)[22], Energy-Efficient Compressive Sensing-based clustering
Routing (EEPC) protocol [23]. Table 2 shows the simulation settings
Table 2. Simulation settings

72
Energy Consumption
The performance of the proposed TCCS-Distributed comprises of the clustering-based
optimization model for improving the overall performance of the WSN environment. In WSN
due to higher node mobility energy consumption is higher to reduce energy consumption
proposed TCCS-Distributed model uses the optimization model. In table 3 the performance of the
proposed TCCS-Distributed model is compared with the CDAS, EEPC and TCCS-Centralized.
Table 3. Energy consumption calculation(J)
Nodes CDAS EEPC TCCS-Centralized TCCS-Distributed
0 0 0 0 0
20 5.127 3.21 2.01 1.08
40 9.178 7.65 4.20 2.16
60 15.46 10.78 5.19 4.11
80 21.45 14.85 8.16 6.25
100 26.97 17.93 11.85 8.04
The energy consumption for varying time for the WSN environment is presented in figure 4.
Energy consumption defines the amount of energy consumed by the network for different time
instances in the network. Figure 5 shows the energy consumption calculations of the proposed
method. The network's energy consumption is simulated and analysed for nodes 0, 20, 40, 60, 80,
and 100. The energy consumption of the convention CDAS, EEPC and TCCS-Centralized is
significantly higher than the proposed TCCS-Distributed. Initially, for time energy consumption
of all nodes are equal to zero.At different nodes 0,20,40,60,80 and 100energy consumption of the
proposed TCCS-Distributed us measured as 2.47, 4.78, 6.78, 9.12 and 11.11 (J) respectively.
Figure 5. Energy Consumption Calculation
However, the energy consumption of conventional CDAS, EEPC and TCCS-Centralized exhibits
higher energy consumption in the WSN network. The analysis of the simulation results expressed
that proposed TCCS-Distributed exhibits ~ 6% reduced energy consumption compared with the
conventional CDAS, EEPC and TCCS-Centralized.

73
Energy Efficiency
Energy efficiency defines the amount of energy remains in the network upon the transmission of
data in the network. As WSN environment comprises of the higher node mobility and energy
consumption level to improve energy efficiency of the WSN network proposed TCCS-
Distributed model uses the optimization model. In table 4 the performance of the proposed
TCCS-Distributed model is compared with the CDAS, EEPC and TCCS-Centralized.
Table 4. Energy efficiency calculation (J)
0 0 0 0 0
20 9.24 13.54 15.11 17.22
40 23.46 33.48 39.64 42.19
60 41.89 49.13 53.89 56.11
80 59.17 64.88 71.45 75.23
100 74.86 82.98 88.67 91.04
Energy efficiency defines the amount of energy remains in the network upon the transmission of
data in the network. As WSN environment comprises of the higher node mobility and energy
consumption level to improve energy efficiency of the WSN network proposed TCCS-
Distributed model uses the optimization model. In figure 5 the performance of the proposed
TCCS-Distributed model is compared with the CDAS, EEPC and TCCS-Centralized. Figure 6
shows energy efficiency calculation.
.
Figure 6. Energy Efficiency Calculation
The simulation analysis of the energy efficiency of the network is evaluated for varying nodes
0,20,40,60,80 and 100. The energy efficiency of the convention CDAS, EEPC and TCCS-
Centralized is significantly minimal than the proposed TCCS-Distributed. Initially, for time 0ms
energy efficiency of all nodes are equal to zero. For the different nodes 0,20,40,60,80 and 100
energy efficiency of the proposed TCCS-Distributed us measured as 19.78, 32.14, 56.47, 78.14
and 89.12(J) respectively. However, the energy efficiency of conventional CDAS, EEPC and
TCCS-Centralized exhibits minimal energy efficiency in the WSN network. The analysis of the
simulation results expressed that proposed TCCS-Distributed exhibits ~ 2% improved energy
efficiency compared with the conventional CDAS, EEPC and TCCS-Centralized.

74
Packet Delivery Ratio
The PDR defines the number of packets received to the total packets transmitted in the network.
The analysis stated that the PDR values in the network are evaluated based on the evaluation of
the proposed and conventional methods in the network as illustrated in the table 5.
Table 5. Packet delivery ratio calculation
0 0 0 0 0
20 17.32 24.55 28.91 32.63
40 32.67 35.91 38.01 42.03
60 41.04 48.26 52.31 50.17
80 63.93 67.94 69.03 71.93
100 80.56 89.02 92.34 93.93
The PDR value of the proposed TCCS-Distributed evaluated for the existing CDAS, EEPC and
TCCS-Centralized. The PDR analysis of the node parameters in the network are presented in
figure 7.
Figure 7. Packet Delivery Ratio Calculation
The simulation analysis of the packet delivery ratio network is evaluated for varying nodes
0,20,40,60,80 and 100. The packet delivery ratio of the convention CDAS, EEPC and TCCS-
Centralized is minimal than the proposed TCCS-Distributed. Initially, for time 0ms packet
delivery ratio of all nodes is equal to zero. At different nodes 0,20,40,60,80 and 100 packet
delivery ratios of the proposed TCCS-Distributed is measured as 22.14, 37.43, 53.86, 76.47 and
94.26 respectively. However, the packet delivery ratio of conventional CDAS, EEPC and TCCS-
Centralized exhibits minimal value in the WSN network. The analysis of the simulation results
expressed that proposed TCCS-Distributed exhibits ~ 3% higher packet delivery ratio compared
with the conventional CDAS, EEPC and TCCS-Centralized.
Packet Loss
The number of packets dropped or missed in the WSN network is presented as shown in table 6.
The comparative analysis of the packet drops for the proposed TCCS-Distributed and the existing
CDAS, EEPC and TCCS-Centralized are presented.

75
Table 6. Packet loss calculation
Nodes CDAS EEPC TCCS-Centralized TCCS-
Distributed
0 0 0 0 0
20 89 76 59 31
40 143 137 147 98
60 221 226 204 149
80 349 318 269 223
100 450 396 320 282
The comparative analysis of the proposed TCCS-Distributed with the conventional technique for
varying time instances are presented in figure 8. The analysis of the packet drop exhibits the
proposed TCCS-Distributed is minimal.
Figure 8. Packet Loss Calculation
The simulation analysis of the packet drop network is evaluated for varying nodes0,20,40,60,80
and 100. The packet drop of the convention CDAS, EEPC and TCCS-Centralized is higher than
the proposed TCCS-Distributed. Initially, for time 0ms packet drop of all nodes are equal to zero.
At varying nodes 0,20,40,60,80 and 100 packet drop of the proposed TCCS-Distributed is
measured as 9, 17, 22, 36 and 46 respectively. However, the packet drop of conventional CDAS,
EEPC and TCCS-Centralized exhibits higher packet drop value in the WSN network. The
analysis of the simulation results expressed that proposed TCCS-Distributed exhibits ~ 5%
minimal packet drop ratio compared with the conventional CDAS, EEPC and TCCS-Centralized.
End to End Delay
The time by the network for transmission of data from source node destination node is defined as
the end-to-end delay. The efficient communication network should have minimal end-to-end
delay in the WSN. In table 7 the measured End-to-End delay for the proposed TCCS-Distributed
is presented comparatively with conventional CDAS, EEPC and TCCS-Centralized.

76
Table 7. End to end delay calculation (ms)
0 0 0 0 0
20 42.18 31.82 15.48 9.22
40 85.19 61.22 43.71 29.34
60 133.97 96.41 79.12 63.85
80 185.17 146.23 134.96 113.47
100 245.94 193.45 176.68 146.72
The end-to-end delay for varying time in WSN environment is presented in figure 9. As the end-
to-end delay involved in computation of the data transmission time between source to destination
node in the WSN environment.
Figure 9. End to End Delay Calculation
The simulation analysis of the end-to-end delay network is evaluated for time varying nodes
0,20,40,60,80 and 100. The energy efficiency of the convention CDAS, EEPC and TCCS-
Centralized is higher than the proposed TCCS-Distributed. Initially, for time 0ms end-to-end
delay of all nodes are equal to zero. The different node 0,20,40,60,80 and 100 end-to-end delay of
the proposed TCCS-Distributed us measured as 33.26, 61.24, 92.48, 135.48 and 168.25(ms)
respectively. However, the end-to-end delay of conventional AOMDV and EHO-AOMDV
exhibits higher value in the WSN network. The analysis of the simulation results expressed that
proposed TCCS-Distributed exhibits ~ 3% minimal end-to-end delay compared with the
conventional CDAS, EEPC and TCCS-Centralized.
Throughput
The computation of the network defines the ratio between number of successful data received to
the total number of packets in the network. The throughput measurement of the network is shown
in table 8.
Table 8. Throughput calculation (kbps)
0 0 0 0 0
20 65.22 87.34 98.53 127.39
40 127.49 154.22 187.25 206.33
60 178.39 221.93 267.22 281.24
80 229.57 265.86 298.45 379.23
100 276.89 321.98 357.13 465.51

77
The computation of throughput is comparatively analyzed with the CDAS, EEPC and TCCS-
Centralized. The performance of the proposed TCCS-Distributed for the estimation of the
throughput for successful transmission and reception of data packets in the network is presented
in figure 10.
Figure 10. Proposed method throughput Calculation
The network's throughput is simulated and analyzed for nodes 0, 20, 40, 60, 80, and 100. The
throughput of the convention CDAS, EEPC and TCCS-Centralized is minimal than the proposed
TCCS-Distributed. Initially, for time 0ms packet drop of all nodes are equal to zero. At different
nodes 0,20,40,60,80 and 100packet drop of the proposed TCCS-Distributed is measured as 54.23,
89.47, 134.16, 169.14 and 205.59kbps respectively. However, the throughput of conventional
CDAS, EEPC and TCCS-Centralized exhibits minimal throughput value in the WSN network.
The analysis of the simulation results expressed that proposed TCCS-Distributed exhibits ~ 5%
improved throughput compared with the conventional CDAS, EEPC and TCCS-Centralized.
6. CONCLUSION
In this work, the impact of CS theory and decentralized clustering results in the betterment of
network lifetime and QoS. To achieve this at the initial stage the network construction in done
with formulated manner. Proposed the Energy Aware Talented Clustering with Compressive
Sensing (TCCS). This approach combines principles such as Compressive Sensing (CS),
Connection-based Decentralized Clustering (CDC), Relay node selection, and Multi Objective
Genetic Algorithm (MOGA) (MOGA). CH selection here is done by computing the Euclidean
distance, which considerably reduces the network's energy usage. The Connection-based
Decentralized Clustering model focuses on the number of hops and the message weight employed
in data transport. Selecting the right relay node might help to save energy usage. Finally, Multi
Objective Genetic Algorithm (MOGA) is used for Energy efficiency and reconstruction error.
When compared to the previous methods, the simulation results show that the proposed work
performs better in terms of the calculation of maximum packet delivery ratio of 93.93 percent,
minimum energy consumption of 8.04J, maximum energy efficiency of 91.04 percent, maximum
network throughput of 465.51kbps, minimum packet loss of 282 packets, and minimum delay of
63.82 msec. In future, we will expand this work by creating multi-level trust model for TCCS
method.
CONFLICTS OF INTEREST
The authors declare no conflict of interest

78
REFERENCES
[1] Veeraiah, N., Krishna, B.T. An approach for optimal-secure multi-path routing and intrusion
detection in MANET. Evol. Intel. 15,1313–1327 (2022).
[2] D. J. Dechene, A. El Jardali, M. Luccini, and A. Sauer, A Survey of Clustering Algorithms for
Wireless Sensor Networks ,Computer Communications, Butterworth-Heinemann Newton, MA USA,
October 2007.
[3] Manjeshwar, A., Agarwal, D.P.: TEEN: a routing protocol for enhanced efficiency in wireless sensor
networks. Proc. 15th Int. Parallel and Distributed Processing Symp (2001): 2009-2015.
[4] Xiong, Z., Liveris, A., & Cheng, S. (2004). Distributed source coding for sensor networks. IEEE
Signal Processing Magazine, 21(5), 80–94.
[5] Zahhad, M. A., Amin, O., Farrag, M., & Ali, A. (2015). An energy consumption model for wireless
sensor networks. In IEEE 5th international conference on energy aware computing systems &
applications, Egypt, March 2015.
[6] Veeraiah, N., Krishna, B.T. Trust-aware FuzzyClus-Fuzzy NB: intrusion detection scheme based on
fuzzy clustering and Bayesian rule. Wireless Netw25, 4021–4035 (2019).
[7] Haupt, J., Bajwa, W., Rabbat, M., & Nowak, R. (2008). Compressed sensing for networked data.
Signal Processing Magazine, 25(2), 92–101.
[8] Mehrjoo, S., &Khunjush, F. (2018). Accurate compressive data gathering in wireless sensor networks
using weighted spatio-temporal compressive sensing. Telecommunication Systems, 68(1), 79–88.
[9] Veeraiah N, Krishna BT. Intrusion detection based on piecewise fuzzy C-means clustering and fuzzy
Naïve Bayes rule. Multimedia Research. 2018 Oct;1(1):27-32.
[10] U. Srilakshmi et al.,"A Secure Optimization Routing Algorithm for Mobile Ad Hoc Networks," in
IEEE Access, vol. 10, pp. 14260-14269, 2022, doi: 10.1109/ACCESS.2022.3144679.
[11] Guleria, Kalpna& Verma, Anil & Goyal, Nitin & Sharma, Ajay & Benslimane, Abderrahim& Singh,
Aman. (2021). An Enhanced Energy Proficient Clustering (EEPC) Algorithm for Relay Selection in
Heterogeneous WSNs. Ad Hoc Networks. 116. 102473. 10.1016/j.adhoc.2021.102473.
[12] V. Seedha Devi, T. Ravi, S. Baghavathi Priya, Cluster Based Data Aggregation Scheme for Latency
and Packet Loss Reduction in WSN, Computer Communications, Volume 149,2020,Pages 36-43.
[13] Pakdaman Tirani, Shima & Avokh, Avid. (2018). On the performance of sink placement in WSNs
considering energy-balanced compressive sensing-based data aggregation. Journal of Network and
Computer Applications. 107. 10.1016/j.jnca.2018.01.012.
[14] Senouci, Mustapha & Mellouk, Abdelhamid. (2019). A robust uncertainty-aware cluster-based
deployment approach for WSNs: Coverage, connectivity, and lifespan. Journal of Network and
Computer Applications. 146. 102414. 10.1016/j.jnca.2019.102414.
[15] Alia, Osama. (2017). Dynamic Relocation of Mobile Base Station in Wireless Sensor Networks
Using a Cluster-Based Harmony Search Algorithm. Information Sciences. 385-386. 76-95.
10.1016/j.ins.2016.12.046.
[16] Anis Jari, Avid Avokh, PSO-based sink placement and load-balanced anycast routing in multi-sink
WSNs considering compressive sensing theory, Engineering Applications of Artificial Intelligence,
Volume 100,2021,104164.
[17] Lv, Cuicui & Wang, Qiang & Li, Jia. (2019). Compressive Sensing-based sequential data gathering in
WSNs. Computer Networks. 154. 10.1016/j.comnet.2019.03.004.
[18] Zhang, Ce & Li, Ou & Yang, Yanping & Liu, Guangyi& Tong, Xin. (2019). Energy-Efficient Data
Gathering Algorithm Relying on Compressive Sensing in Lossy WSNs. Measurement. 147. 106875.
10.1016/j.measurement.2019.106875.
[19] G. Yang, M. A. Jan, V. G. Menon, P. G. Shynu, M. M. Aimal and M. D. Alshehri, "A Centralized
Cluster-Based Hierarchical Approach for Green Communication in a Smart Healthcare System," in
IEEE Access, vol. 8, pp. 101464-101475, 2020, doi: 10.1109/ACCESS.2020.2998452.
[20] S. M. M. H. Daneshvar, P. AlikhahAhariMohajer and S. M. Mazinani, "Energy-Efficient Routing in
WSN: A Centralized Cluster-Based Approach via Grey Wolf Optimizer," in IEEE Access, vol. 7, pp.
170019-170031, 2019, doi: 10.1109/ACCESS.2019.2955993.
[21] M. A. Mazaideh and J. Levendovszky, "A multi-hop routing algorithm for WSNs based on
compressive sensing and multiple objective genetic algorithm," in Journal of Communications and
Networks, vol. 23, no. 2, pp. 138-147, April 2021, doi: 10.23919/JCN.2021.000003.
[22] Ghaderi, M.R., TabatabaVakili, V. & Sheikhan, M. Compressive sensing-based energy consumption
model for data gathering techniques in wireless sensor networks. Telecommun Syst77, 83–108 (2021).

79
[23] Q. Wang, D. Lin, P. Yang and Z. Zhang, "An Energy-Efficient Compressive Sensing-Based
Clustering Routing Protocol for WSNs," in IEEE Sensors Journal, vol. 19, no. 10, pp. 3950-3960, 15
May15, 2019, doi: 10.1109/JSEN.2019.2893912.
[24] Zeyu Sun, Rong Tao, NaixueXiong, Xiaoyan Pan, "CS-PLM: Compressive Sensing Data Gathering
Algorithm Based on Packet Loss Matching in Sensor Networks", Wireless Communications and
Mobile Computing, vol. 2018, Article ID 5131949, 12 pages, 2018.
[25] U. Srilakshmi et al., "An Improved Hybrid Secure Multipath Routing Protocol for MANET," in IEEE
Access, vol. 9, pp. 163043-163053, 2021.
[26] N. Veeraiah et al., "Trust Aware Secure Energy Efficient Hybrid Protocol for MANET," in IEEE
Access, vol. 9, pp. 120996-121005, 2021.
[27] N. Veeraiah and B. T. Krishna, "Selfish node detection IDSM based approach using individual master
cluster node," 2018 2nd International Conference on Inventive Systems and Control (ICISC), 2018,
pp. 427-431.
[28] G.Srinivasalu and H.Umadevi, “Cluster Based Routing Using Energy And Distance Aware Multi-
Objective Golden Eagle Optimization In Wireless Sensor Network,” International Journal of
Computer Networks & Communications (IJCNC) Vol.14, No.3, May 2022.
[29] Prakash K Sonwalkarand Vijay Kalmani, “Energy Harvesting Rectenna Design For Enhanced Node
Lifetime In WSNs,” International Journal of Computer Networks & Communications (IJCNC)
Vol.14, No.2, March 2022.

Energy Aware Talented Clustering with Compressive Sensing (TCCS) for Wireless Sensor Networks

More Related Content

Similar to Energy Aware Talented Clustering with Compressive Sensing (TCCS) for Wireless Sensor Networks (20)

More from IJCNCJournal (20)

Recently uploaded (20)

Energy Aware Talented Clustering with Compressive Sensing (TCCS) for Wireless Sensor Networks