International Journal of Reconfigurable and Embedded Systems (IJRES)

Vol. 13, No. 3, November 2024, pp. 774~785

ISSN: 2089-4864, DOI: 10.11591/ijres.v13.i3.pp774-785  774

Energy-efficient clustering and routing using fuzzy k-medoids

and adaptive ranking-based wireless sensor network

Haritha K. Sivaraman, Rangaiah Leburu

Department of Electronics and Communication Engineering, RajaRajeswari College of Engineering, Bangalore, India

Article Info ABSTRACT

Article history: The wireless sensor network (WSN) is a vital component of infrastructure
that is seeing tremendous demand and quick expansion in a variety of
Received Feb 19, 2024 industries, including forestry, airports, healthcare, and the military.
Revised Mar 21, 2024 Increasing network lifetime and reducing power consumption (PC) are now
Accepted Apr 19, 2024 major goals in WSN research. This research proposes a unique energy-
efficient cross-layer WSN design that aims to maximize network lifetime
while maintaining quality of service (QoS) criteria to address these
Keywords: challenges. The research initially utilizes the fuzzy k-medoids (FKMeds)
clustering technique to group sensor nodes (SN) to improve resilience,
ARanEOR scalability, and minimize network traffic. Following that, the hybrid
Cluster head improved grey wolf and ant colony (HIGWAC) optimization approach is
Energy applied to choose cluster heads (CH), minimizing distances, reducing
Fuzzy k-medoids latency, and optimizing energy stability. Finally, data is transmitted through
Routing the shortest pathways using the adaptive ranking-based energy-efficient
Wireless sensor network opportunistic routing (ARanEOR) protocol, which ensures effective and
energy-conserving routing in WSN while dynamically lowering network
overhead. Compared to existing approaches, the proposed method in this
study outperforms them in terms of energy efficiency, latency, and network
longevity.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Haritha K Sivaraman
Department of Electronics and Communication Engineering, RajaRajeswari College of Engineering
Ramohalli Cross, Mysore Rd, Kumbalgodu, Bangalore, Karnataka, India
Email: [email protected]

1. INTRODUCTION
The energy efficient clustering protocol addresses the critical problem of energy consumption (EC)
in wireless sensor network (WSN), where resource-constrained sensor nodes (SN) are critical. It is widely
employed in several applications, WSNs need energy-saving methods to extend the life of their networks [1].
The protocol attempts to cluster SN and optimize the selection of the cluster head (CH) by considering node
density, residual energy, and distance, among other variables. Through the use of efficient clustering
algorithms to enable data aggregation at the cluster level, the protocol reduces the energy needed for data
transmission to a central base station (BS) [2]–[4]. Reducing EC and extending network lifetime are the
primary goals, and they ensure the long-term and dependable operation of WSNs [5].
The CH-based cross-layer routing protocol is a smart solution for major issues in WSNs. The
protocol offers a novel approach to routing by utilizing CH as the main decision-makers and integrating
information from the physical, data link, and network levels for effective routing. CH are critical to data
transmission because they make smart decisions that optimize energy efficiency and minimize congestion
[6]. The cross-layer integration ensures the data transmission paths that use the least amount of energy. CH-
based decision-making at various network layers, which is the protocol's main focus, is anticipated to

Journal homepage: http://ijres.iaescore.com

Int J Reconfigurable & Embedded Syst ISSN: 2089-4864  775

significantly improve WSN performance. The integration of energy efficient clustering and CH-based cross-
layer routing provides a holistic solution to the challenges inherent in WSNs. The protocol aims to strike a
balance between energy conservation and effective data communication. This paper presents an innovative
approach for energy-efficient cross-layer WSN design in response to these demands, to maximize network
lifetime and ensure rigorous quality of service (QoS) standards simultaneously [7]–[9].

2. LITERATURE REVIEW
This section examined some recent studies on energy-efficient clustering and cluster CH-based
cross-layer routing protocols in WSNs. The study being conducted on energy efficiency and carbon reduction
(EECR) in WSNs utilizing fuzzy k-medoids (FKMeds) and an adaptive ranking-based cross-layer routing
protocol would benefit greatly from this review, which will also set theoretical groundwork for future
research in this area. In 2020, Deepa and Suguna [10] examined the routing protocol difficulties and solutions
in WSN. It proposed an optimized QoS-based clustering with multipath routing protocol (OQoS-CMRP) to
enhance network lifespan, data dependability, and energy preservation. It highlights the importance of
balancing energy usage and data quality in QoS-based routing protocols. OQoS-CMRP is compared with
existing protocols. It maximizes network routing and clustering efficiency in WSNs [11], [12]. In 2020,
Doostali and Babamir [13] discussed the WSN EC issues and provided an energy-efficient CH selection
technique. It focuses attention on the shortcomings of current methods and presents clustering as a distributed
energy management strategy. The technique, based on learning automata and sleep-awake mechanism,
enhances network endurance and energy usage compared to previous methods. Mehbodniya et al. [14] in
2022, introduced the energy-aware proportional fairness multi-user routing (EPFMR) architecture in WSN,
an energy-efficient routing system. The first phase in EPFMR framework's design to lower the rate of EC is
the request time submitted for route finding. The greedy instance fair method (GIFM) was used to quantify
energy on multi-user route paths when a route path for packet flow was found. Throughput is increased by
GIFM in EPFMR, which creates node-dependent, energy-efficient localized route paths.

3. SYSTEM MODEL
An advanced network architecture known as WSNs is intended to continuously observe and collect
information on the physical condition of the area in which they are located. The BS and many SNs are two
main parts of this infrastructure that are usually available. The BS acts as the hub or main control unit for the
WSN. It serves as an interface between end users or data analysis systems and the field-deployed
SN [15], [16].

3.1. Network model

WSN clustering splits sensors into clusters to use as little energy as possible. The CH receives
sensory input from common nodes that regularly scan the surroundings. Among the common nodes, the CH
node is perpetually selected. An essential role of CH is to collect data from each cluster node and forward it
to BS. Grouping helps to prevent direct connection between sensors and receivers. Figure 1 is an illustration
of WSN system model.

3.2. Concept of energy

If the threshold distance ′𝑑𝑡ℎ ′ is larger than the value of ‘𝑑𝑝 ′, then node's energy is precisely
proportionate to propagation distance, '𝑑𝑝 ′. The (1) represents the total energy required by each node to
transport 𝑀-bit data packet.

𝐸𝑇 (𝑀, 𝑑𝑝 ) = 𝐸𝑅 (𝑀) + 𝜀𝑎𝑚 × 𝑀 (1)

Where, 𝐸𝑇 indicates energy required for data transmission; 𝐸𝑅 represents the energy consumed for receiving
data; 𝜀𝑎𝑚 denotes energy used for amplification; 𝑀 is number of bits.
The EC in receiver end is given in (2).

𝐸𝑅 (𝑀) = 𝐸𝑑𝑖𝑠 × 𝑀 (2)

Where, 𝐸𝑑𝑖𝑠 indicates energy dissipation per bit.

The amplification energy can be expressed as (3).

Energy-efficient clustering and routing using fuzzy k-medoids … (Haritha K. Sivaraman)

776  ISSN: 2089-4864

𝜀𝑓𝑠 × 𝑑𝑝 2 , 𝑤ℎ𝑒𝑛 𝑑𝑝 ≤ 𝑑𝑡ℎ

𝜀𝑎𝑚 = { (3)
𝜀𝑚𝑝 × 𝑑𝑝 2 , 𝑤ℎ𝑒𝑛 𝑑𝑝 > 𝑑𝑡ℎ

Where, 𝜀𝑓𝑠 represents energy utilized for amplification in free space; 𝜀𝑚𝑝 indicates amplification energy used
in multi path [17]-[19].
The overall loss of energy for WSN is represented as 𝐸𝑇𝑜𝑡𝑎𝑙 and the expression of total energy is
given in (4).

𝐸𝑇𝑜𝑡𝑎𝑙 = 𝐸𝑇 + 𝐸𝑅 (4)

Base Station

Sensor Node

Cluster Head

Figure 1. General structure of WSN

4. PROPOSED METHOD
This section discusses in detail the proposed EECR using FKMeds and adaptive ranking-based cross
layer routing protocol for WSN. The proposed method has three methods: initially, FKMeds method is used
for clustering to achieve robustness, scalability, and reduced network traffic SN and it is briefly discussed in
section 4.1. Secondly, hybrid improved grey wolf and ant colony (HIGWAC) optimization technique is
proposed to select CH through distance reduction, energy stabilization, and latency minimization between
nodes and this technique is covered in section 4.2. Finally, adaptive ranking-based energy-efficient
opportunistic routing (ARanEOR) protocol is applied to transmit data across the shortest channels for
efficient and energy-preserving routing in WSN and dynamically lowering network overhead and this
approach is discussed in section 4.3.

4.1. Clustering using FKMeds

In this study, the SN are grouped using clustering approach called FKMeds method, which
calculates the distance criteria by computing the values of cluster center of data. The updated membership
matrix 𝑀𝑢 is first determined using the fuzzy c means (FCM) approach in the FKMeds in order to obtain the
medoid value [20]-[23]. The index data from each cluster with the greatest membership value is then used to
choose the medoid. The FKMeds objective function is represented in (5).

𝐹𝑜𝑏𝑗 = ∑𝑚 𝑛 2 𝑟
𝑖=1 ∑𝑘=1(𝑑 (𝑐𝑘 , 𝑥𝑖 )(𝛿𝑖𝑘 ) ) (5)

Where, 𝐹𝑜𝑏𝑗 is objective function in 𝑡-th iteration; 𝜔𝑖𝑘 represents degree in 𝑀𝑢 membership matrix; 𝑟 denotes
fuzzy rank ( 𝑟 ≥ 2); 𝑑(𝑐𝑘 , 𝑥𝑖 ) indicates distance between 𝑖 th data value and the 𝑘th cluster center.
The initial membership degree value ‘𝛿𝑖𝑘 ′ in 𝑀𝑢 membership matrix is created using the (6) and (7)
when using FKMeds approach for cluster center computations.

𝑀𝑢 = [𝛿𝑖𝑘 ]𝑚×𝑛 ,∑𝑐𝑘=1 𝜔𝑖𝑘 =1, 1≤𝑖≤𝑛 (6)

Int J Reconfigurable & Embedded Syst, Vol. 13, No. 3, November 2024: 774-785
Int J Reconfigurable & Embedded Syst ISSN: 2089-4864  777

𝛿𝑖𝑘 = [0, 1], 𝑖 = 1, 2, … , 𝑚; 𝑘 = 1, 2, 3, 4, … , 𝑐 (7)

Each iteration updated the membership matrix, represented by 𝛿𝑖𝑘 , which is calculated using (8).
−1
[𝑑 2 (𝑐𝑘 ,𝑥𝑖 )]𝑟−1
𝛿𝑖𝑘 = −1 (8)
∑𝑐𝑗=1[𝑑 2 (𝑐𝑗 ,𝑥𝑖 )]𝑟−1

The cluster center is computed using the (9) once the 𝑀𝑢 membership matrix has been obtained.

∑𝑚 𝑟
𝑖=1(𝜔𝑖𝑘 ) 𝑦𝑖
𝐶𝑘 = ∑𝑚 𝑟 (9)
𝑖=1(𝜔𝑖𝑘 )

4.2. Cluster head selection using hybrid improved grey wolf and ant colony
In WSN, CH is a node that serves as coordinator for other cluster nodes. It is responsible for
collecting data from each cluster node and forwarding it to BS. Data from other nodes within the cluster is
gathered by the CH. The data may include of any measurement made by the sensors. The CH may compile
and analyze the gathered data before forwarding it to minimize the volume of information sent. The CH
sends the processed data to BS, the WSN's central hub. This might be accomplished directly or in a multi-hop
manner via other CH. A crucial component of WSNs is CH selection, which has an impact on energy
efficiency and the lifespan of network. Although existing CH selection methods have a great deal to offer
WSNs, they frequently have issues with node delay, energy stability, and distance reduction. To overcome
these limitations, this study introduces HIGWAC optimization approach which is a combination of improved
grey wolf optimization (IGWO) and ant colony optimization (ACO) algorithm [24], [25].

4.2.1. Improved grey wolf optimization

The information on optimizing CH choice process to maximize network lifetime is offered to IGWO
algorithm. CH selection is carried out by BS using IGWO algorithm to eliminate unpredictability inherent.
Based on fitness values, all SNs are separated into four subgroups, roughly, of which sixteen SN locations are
designated as fixed to accommodate multi pathways. The method is stated in terms of rounds. An SN's fitness
value is calculated using its distance from BS and leftover energy.

𝐸𝑟𝑒𝑠 𝐷𝑚𝑎𝑥 −𝐷
0.8 × ( ) + 0.2 × ( ) , 𝐸𝑟𝑒𝑠 < 0
𝐸𝑖𝑛𝑖𝑡 𝐷𝑚𝑎𝑥 −𝐷𝑚𝑖𝑛
𝐹𝑖𝑡 = { 𝐸𝑟𝑒𝑠 𝐷𝑚𝑎𝑥 −𝐷
(10)
0.2 × ( ) + 0.8 × ( ) , 𝐸𝑟𝑒𝑠 < 𝑑𝑡ℎ
𝐸𝑖𝑛𝑖𝑡 𝐷𝑚𝑎𝑥 −𝐷𝑚𝑖𝑛

Where, 𝐸𝑟𝑒𝑠 represents residual energy at SN after every cycle, 𝐸𝑖𝑛𝑖𝑡 indicates initial energy of SN, 𝐷 is
distance between SN and BS, 𝐷𝑚𝑖𝑛 and 𝐷𝑚𝑎𝑥 are minimum and maximum distance between BS and SN
respectively.
The initial position of BS is calculated using (11).

⃗⃗⃗⃗⃗⃗
𝑃𝐶𝐻 = |𝑊𝛼 ⃗⃗⃗
𝑃𝛼 + 𝑊𝛽 ⃗⃗⃗⃗
𝑃𝛽 + 𝑊𝛾 ⃗⃗⃗
𝑃𝛾 | (11)

Where, 𝑊𝛼 , 𝑊𝛽 , 𝑊𝛾 are initial weights and it can be expressed as (12).

𝐹𝑖𝑡∝ 𝐹𝑖𝑡𝛽 𝐹𝑖𝑡𝛾

𝑊𝛼 = , 𝑊𝛽 = , 𝑊𝛼 = (12)
𝐹𝑖𝑡∝ +𝐹𝑖𝑡𝛽 +𝐹𝑖𝑡𝛾 𝐹𝑖𝑡∝ +𝐹𝑖𝑡𝛽 +𝐹𝑖𝑡𝛾 𝐹𝑖𝑡∝ +𝐹𝑖𝑡𝛽 +𝐹𝑖𝑡𝛾

Where, 𝐹𝑖𝑡∝ , 𝐹𝑖𝑡𝛽 , and 𝐹𝑖𝑡𝛽 indicates first three optimal values of SNs. The IGWO technique is utilized to
improve global search capabilities. The weights 𝑊𝛼 , 𝑊𝛽 , and 𝑊𝛾 are dynamically updated with the help of the
⃗⃗⃗ 𝐶 at 𝑖 th iteration, the weights are computed as (13) to (15).
vectors 𝐷,

⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝐷 𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1
𝛼 ×𝐶𝛼
𝑊𝛼𝑖+1 = ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 +𝐷 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1
(13)
𝐷𝛼 ×𝐶𝛼 +𝐷𝛽 ×𝐶𝛽 𝛾 ×𝐶𝛾

⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝐷𝛽 𝑖+1 ×𝐶 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1
𝛽
𝑊𝛽𝑖+1 = ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗ 𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1
(14)
𝐷𝛼 ×𝐶 𝛼 +𝐷𝛽 ×𝐶𝛽 +𝐷𝛾 ×𝐶𝛾

Energy-efficient clustering and routing using fuzzy k-medoids … (Haritha K. Sivaraman)

778  ISSN: 2089-4864

⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝐷𝛾𝑖+1 ×𝐶 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1
𝛾
𝑊𝛾𝑖+1 = ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗ 𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1 ⃗⃗⃗⃗⃗⃗⃗⃗⃗⃗
𝑖+1
(15)
𝐷𝛼 ×𝐶 𝛼 +𝐷𝛽 ×𝐶𝛽 +𝐷𝛾 ×𝐶𝛾

During CH selection process, α, β, and 𝛾 wolves are used to identify the position of CH while other SNs
compute their distances in relation to BS. In (𝑖 + 1)th iteration, the revised location of SN is calculated as
(16).

⃗⃗⃗⃗⃗⃗⃗⃗
𝑃𝑖+1 = ⃗⃗⃗⃗⃗⃗
𝑖
𝑃𝐶𝐻 ⃗ ×𝐶
− 𝐷 (16)

⃗⃗ × ⃗⃗⃗
Where, 𝐶 indicates convergence vector and it is calculated as 𝐶 = 2 ∝ ⃗⃗ , The CH location from the
𝑟1 −∝
th ⃗⃗⃗⃗⃗⃗
𝑖
previous iteration, or 𝑖 iteration is represented by 𝑃𝐶𝐻 .

4.2.2. Ant colony optimization

The probability algorithm for identifying optimization pathways is called ACO. The ant system has
three components: Initialization is the beginning phase. Every ant needs a taboo table to keep track of notes
that are transferred throughout the procedure. Reduce the starting value of the pheromone on either side. Set
the initial value of taboo table to the ant's notice that it is l in length. Set the initial value of pheromone that
the ant releases at each side to zero. Step two is building the route. Ants use a specific probability to decide
the CH’s future location. Probabilities are computed using the (17).
𝛼 𝛽
[𝜋𝑖𝑗 (𝑡)] [𝜂𝑖𝑗]
𝑖𝑓 𝑗𝜖𝑎𝑙𝑙𝑜𝑤𝑒𝑑
𝑃𝑖𝑗 (𝑡) = {∑𝑤𝜖𝑎𝑙𝑙𝑜𝑤𝑒𝑑[𝜋𝑖𝑗 (𝑡)]𝛼[𝜂𝑖𝑗]𝛽 (17)
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

In the (17) mentioned above 𝛼 is a weight calculation of pheromone and indicates the probability of
selecting nodes at time t from CH 𝑖 to CH 𝑗. The pheromone that helps ants choose which node to target next
is more efficient the higher 𝛼. The inspiration factor, denoted by 𝛽, plays a crucial role in the probability
calculation process. A higher value of 𝛽 indicates that the ants will have an easier time selecting the next
target CH. A list of CHs not on the ant's forbidden table is permitted. The pheromone's role comes in third.
According to ACO model, the pheromone is updated as (18) after every ant discovers.

𝜏(𝑡 + 1) = 𝜌 ∙ 𝜏𝑖𝑗 (𝑡) + ∆𝜏𝑖𝑗 (𝑡, 𝑡 + 1) (18)

Where, 𝜏𝑖𝑗 (𝑡) indicates the pheromone of edge 𝑖𝑗 at time 𝑡. 𝜌 defines maintenance factor. In (18) 𝜌, 𝛼, and 𝛽
are the optimal parameters. These parameters are used to enhance the performance of model by tuning the
parameters. The ρ and 𝛼 are robust at 0.6 and 0.2, respectively. When more precise solutions are required,
lower values of parameter β are more suitable. It ranges from 6 to 12.

4.2.3. Hybrid improved grey wolf and ant colony optimization

A hybrid technique that combines the advantages of IGWO and ACO can be very beneficial for CH
selection in WSNs, especially when it comes to stability, latency minimization, and distance reduction. To
improve the efficiency of CH selection in WSN, the ACO is fused with IGWO. Therefore, the adaptability of
ACO's pathfinding allows it to adapt to changes in the network, while the effective optimization of IGWO
ensures near-optimal performance. To select the optimal CH, the optimal parameter values 𝛼 = 0.2, 𝛽 = 6,
and 𝜌 = 0.6 of ACO is applied to the weight parameters 𝑊𝛼 , 𝑊𝛽 , and 𝑊𝛾 of IGWO respectively. The optimal
CH of HIGWAC can be expressed as (19).

⃗⃗⃗⃗⃗⃗
𝑃𝐶𝐻 = |(0.2 × ⃗⃗⃗
𝑃𝛼 ) + (6 × ⃗⃗⃗⃗
𝑃𝛽 ) + (0.6 × ⃗⃗⃗
𝑃𝛾 )| (19)

By applying optimal parameters of ACO with IGWO will tune the optimal CH effectively.

4.3. Routing using ARanEOR protocol

ARanEOR protocol uses node ranking based on energy and distance for adaptive route selection.
The WSN routing protocol known as ARanEOR grades forwarder nodes (FNs) and CH using an adaptive
ranking mechanism and employs an opportunistic forwarding technique to maximize EC for each SN. The
major purpose of this mechanism is to dynamically rank nodes, and then utilize that ranking to determine FN
in ARanEOR. The two main components that influence dynamic ranking are node's location and residual

Int J Reconfigurable & Embedded Syst, Vol. 13, No. 3, November 2024: 774-785
Int J Reconfigurable & Embedded Syst ISSN: 2089-4864  779

energy. ARanEOR selects collection of possible nodes and creates a ranking table to determine which node
should be the forwarder. Therefore, by employing refined selection method that is covered in the sections
below, ARanEOR seeks to prevent any needless energy loss, which might increase WSN efficiency and
lower EC.

4.3.1. Architecture and layout

The FN is chosen among source node's neighbors depending on variables like residual energy of
node and distance between nodes. The value of parameter determines how CH and all neighbors of source
node are arranged or given priority in table given below. The process of selecting FN, which is a node nearest
to source node and has the maximum residual energy, is shown in Figure 2.

Cluster 3 Cluster 4
20
22 CH
Sink Node
25

CH 24
18 16 23 27
26

17 21 29

28 30
19

Cluster 2 Cluster 1
CH
7
10 2
15 CH 5
9 12
8 3 4
11
6
14

1
13

Sensor Node Sensor links

Source Node Opportunistic Route

Forward Node Data Relay Route

Inter network Route
Opportunistic Relay

Figure 2. Volunteer routing scheme with opportunistic approach using adaptive node ranking

4.3.2. Routing table

The ARanEOR routing table differs from the one used in conventional opportunistic routing
networks. This change results from the next forwarder node in ARanEOR being chosen depending on rank.
As demonstrated in Figure 2, the next forwarder node in opportunistic and relay routes will be chosen with
the use of a rating table. The example that follows illustrates this:
− Example 1: FN section for Node ID 4 next hop in cluster 1. Table 1 indicates that node with ID 3 has
the highest rank based on residual energy and distance from source node. The next FN is selected from
source node.
− Example 2: Selecting next hop and locating FN in cluster 1 for ID 2. The node with node ID 5 has
highest rank, as indicated by Table 2, based on its distance and residual energy from source node. It has
been selected as FN by relay node with ID 3.

Table 1. Neighbor node ranking from source node

ID of node Range Residual energy (%) Distance
3 1 90 0.45 m
1 3 88 0.78 m
2 2 89 0.49 m
CH 4 85 1.32 m

Energy-efficient clustering and routing using fuzzy k-medoids … (Haritha K. Sivaraman)

780  ISSN: 2089-4864

Table 2. Adjacent nodes ranked by the preceding forwarder node

Node ID Range Residual energy (%) Distance
5 2 87 0.81m
6 1 89 0.69m
CH 3 85 1.2m

4.3.3. Algorithm of ARanEOR

The adaptive node ranking method determines a rank of node based on its distance parameter and
residual energy level. The FNs are chosen based on determined rank. Two factors are recorded for each
communication from each node: distance between nodes (𝐷(𝛿𝑖 , 𝛿𝑗 ) and net residual energy of nodes
(𝐸𝑟𝑒𝑠 (𝛿𝑖 )). Each node's net residual energy is calculated from its residual energy following packet delivery.
Subsequently, among the candidate nodes, the node with the highest net residual energy is taken into
consideration for top ranking, and so on. The nodes with lowest rank are selected to become CH. Once CH
has been located, nodes and their positions are determined with respect to source node. Consequently, the
ranking system helps establish a relay channel for data packets, Algorithm 1 depicts the details of
optimization. Furthermore, the rankings are changed based on updated remaining energy after every relay
round.

Algorithm 1. ARanEOR - Adaptive ranking for FNs

1: Inputs:
𝐷(𝛿𝑖 , 𝛿𝑗 ), 𝐸𝑟𝑒𝑠 (𝛿𝑖 ), 𝑀𝑟 , 𝑀𝑡 , 𝑃𝑡𝑥𝑝 = 0.9 𝐽/𝑏𝑖𝑡, 𝐸𝑖𝑛𝑖𝑡 (𝛿𝑖 ), 𝑁𝑠 , 𝑃𝑟𝑥𝑝 = 0.8 𝐽/𝑏𝑖𝑡
2: Outputs:
𝐸𝑚𝑎𝑥 (𝛿𝑖 ), 𝐷𝑁 (𝛿𝑖 , 𝛿𝑘 )
3: Step 1: Calculate residual energy for each node
4: for i = 1 to N do
5: 𝐸𝑟𝑒𝑠 (𝛿𝑖 ) = 𝑀 − ((𝑀𝑟 ∗ 𝑃𝑟𝑥𝑝 ) + (𝑀𝑡 ∗ 𝑃𝑡𝑥𝑝 )) ;
6: 𝐸𝑚𝑎𝑥 (𝛿𝑖 ) = 𝑚𝑎𝑥(𝐸𝑟𝑒𝑠 (𝛿𝑖 ), 𝐸𝑟𝑒𝑠 (𝛿𝑖 − 1))
7: end for
8: Step 2: Identification of CH
9: if 𝐸𝑚𝑎𝑥 (𝛿𝑘 ) 𝑜𝑐𝑐𝑢𝑟𝑠; 𝑘 𝜖 {1,2, … 𝑀} then
10: 𝛿𝑘 𝜖 𝐶𝐻
11: end if
12: Step 3: Calculate distances between present and forwarder node
13: ∀ 𝛿𝑖 𝜖 𝑁𝑠 𝑎𝑛𝑑 𝛿𝑗 𝜖 𝐹𝑛
2 2
14: 𝐷(𝛿𝑖 , 𝛿𝑗 ) = √(𝑥𝑖 − 𝑥𝑗 ) + (𝑦𝑖 − 𝑦𝑗 )
15: Step 4: Update routing table
16: return

5. RESULTS AND DISCUSSION

This section presents simulation setup and analyses the performance of suggested strategy based on
significant factor results. To evaluate proposed procedure, experiments are carried out in simulation
environment that can imitate wireless network data transmission circumstances in almost real time. For this
purpose, the Python platform has been chosen for implementation.

5.1. Evaluation parameters and criteria

The proposed protocol was implemented on a network of 50 to 200 nodes in steps. On the other
hand, several metrics are used to assess an effectiveness of suggested technique, including throughput, power
consumption (PC), error estimate, EC, packet delivery rate (PDR), packet loss rate (PLR), and end-to-end
latency:
− MSR: it defines the successful completion of whole message transfer process from source node to CH.
− PDR: it is the percentage of packets that are delivered successfully out of all packets that are transmitted
across network.
− PLR: the proportion of data packets in communication system or computer network that are not
successful in reaching their intended destination.
− EC: a typical sensor node gathers data and sends it to CH which uses energy among other two unique
responsibilities. In a similar vein, energy is used by CH when receiving, combining, and transmitting
data.

Int J Reconfigurable & Embedded Syst, Vol. 13, No. 3, November 2024: 774-785
Int J Reconfigurable & Embedded Syst ISSN: 2089-4864  781

− End-end-delay: the total of all the delays brought on by each participating node in a particular
transmission is known as the end-to-end delay. The calculation of node delay is based on the time taken
for communication to begin and end.
− Throughput: throughput measures the rate of successful message delivery within the network over a
specific period.
− PC: it is the total amount of electrical energy used by each sensor node or by network as a whole when
it is in use.
− Network lifetime: network lifetime in the context of WSNs refers to the duration over which the
network can effectively operate before a significant number of sensor nodes become depleted of their
energy resources or fail.
− Error estimation: it refers to the process of assessing or quantifying the accuracy or correctness of a
measurement, calculation, prediction, or any other output in comparison to a known or expected value.

5.2. Evaluation of proposed method with existing techniques

The proposed system is evaluated for effectiveness by comparing it with current methods such as
GIFM, stator flux-oriented (SFO), dual cluster head k-means (DCK-LEACH), and monte carlo gradient sign
attack (MGSA). A simulation is used to assess each protocol's performance, and several factors related to
number of nodes are compared. The evaluation results with 100 nodes are shown in Table 3.

Table 3. Evaluation of Proposed and existing method for 100 nodes

Model ARanEOR (proposed) GIFM SFO DCK-LEACH MGSA
Message success rate 230 185 215 220 195
Packet delivery ratio 93 87 85 90 89
PLR 42 57 50 47 45
EC 150 167 180 161 157
End-to-end delay 62 70 65 74 71
Throughput 215 190 195 205 210
PC 120 127 134 135 141
Network lifetime 220 205 198 200 215
Error estimation 125 132 140 131 135

From the aforementioned results, the proposed algorithm demonstrates magnetic shielding room
(MSR) ranging 240 bits/s for 50 nodes, 230 bits/s for 100 nodes, 225 bits/s for 150 nodes, and 220 bits/s for
200 nodes. The MSR range of proposed method is better than all other existing methods. This result shows
that the proposed protocol is more efficient in cross layer routing for WSN. The graphical representation of
performance analysis of MSR for various number of nodes is displayed in Figure 3.

Figure 3. Proposed routing protocol performance variation by means of MSR compared with other protocols

Energy-efficient clustering and routing using fuzzy k-medoids … (Haritha K. Sivaraman)

782  ISSN: 2089-4864

The proposed protocol has recorded the PDR for 95% for 50 nodes, 93% for 100 nodes, 90% for
150 nodes, and 88% for 200 nodes. Out of all the methods currently in use, the suggested approach has the
highest percentage of PDR. This outcome demonstrates how much more effective the suggested technique is
at cross-layer routing for WSN. Figure 4 shows the graphical depiction of the PDR performance analysis for
different numbers of nodes.

Figure 4. Comparison of suggested routing protocol's performance variation using MSR against other
protocols

The suggested approach has recorded the PLR values of 35, 42, 45, and 40 for the number of nodes
50, 100, 150, and 200 respectively. This achieved is less than other existing approaches. This least PLR value
of proposed technique proves that, the method having less loss rate and it is well suited for cross layer routing
for WSN when compared to existing methods. The analysis of PLR and EC are shown in Figure 5, where
Figure 5(a) shows the PLR and Figure 5(b) shows the EC.

(a) (b)

Figure 5. Analysis of proposed protocol with other protocols over (a) PLR and (b) EC

The energy consumed for 50, 100, 150, and 200 nodes of proposed method are recorded as 145 mJ,
150 mJ, 155 mJ, and 150 mJ respectively. The recorded EC of developed approach is less when compared to
other techniques. A model with less EC will well suited for WSN routing. Therefore, the proposed routing
protocol is suitable for cross layer routing. For the numbers of nodes 50, 100, 150, and 200, respectively, the

Int J Reconfigurable & Embedded Syst, Vol. 13, No. 3, November 2024: 774-785
Int J Reconfigurable & Embedded Syst ISSN: 2089-4864  783

recommended method has recorded delay values of 68, 62, 58, and 45 seconds. This is less than other
methods now in use. When the suggested strategy is compared to the current techniques, its lowest end-end-
delay value indicates that it has a lowest delay and is a good fit for cross-layer routing in WSN. This analysis
is displayed in Figure 6. Figure 6(a) shows the graphical analysis of end-to-end analysis and Figure 6(b)
shows the throughput of proposed and current protocols

(a) (b)

Figure 6. Various node counts at model (a) end-end-delay analysis and (b) throughput of proposed and
current protocols

The proposed approach recorded network lifetime of 225 seconds with 50 nodes, 220 seconds with
100 nodes, 215 seconds with 150 nodes, and 210 seconds with 200 nodes. The lifetime of network serves as
one of the important metrics for evaluating the overall performance of the model. These recorded values
surpass those of all other existing methods, affirming the efficiency of the suggested protocol for WSN
routing. Refer to Figure 7 for graphical analysis of lifetime of network. Figure 7(a) shows the graphical
analysis of lifetime of network and Figure 7(b) shows the error estimation of different models.

(a) (b)

Figure 7. Analysis of proposed and other techniques over in (a) network life time and (b) error estimation.

6. CONCLUSION
This study presents a novel method for energy-efficient clustering and routing in WSNs. We achieve
this by using the following techniques: FKMeds clustering, HIGWAC optimization for CH, and ARanEOR
protocol for data transmission. This work has mainly focused on improving energy efficiency, ensuring QoS

Energy-efficient clustering and routing using fuzzy k-medoids … (Haritha K. Sivaraman)

784  ISSN: 2089-4864

factors in WSNs, and extending the network lifetime. By using FKMeds clustering, this study able to cut
network traffic, enhance scalability, and improve robustness, all of which improved operational efficiency.
The adoption of HIGWAC optimization for CH selection emphasized energy stability, distance reduction,
and latency minimization among network nodes, adding greatly to energy efficiency and network stability.
Furthermore, by enabling data to be sent along the most energy-efficient paths, the ARanEOR protocol
dynamically decreased network overhead and guaranteed effective energy use. The suggested approach was
assessed using a range of performance measures, including message success rate, error estimation,
throughput, PDR, PC, EC, PLR, and end-to-end latency. The proposed approach to WSN optimization is
successful, as evidenced by the higher performance it showed in comparison to existing approaches in terms
of EC, throughput, PC, network lifespan, and error prediction.

REFERENCES
[1] A. M. Khedr, P. P. Raj, and A. Al Ali, “An energy-efficient data acquisition technique for hierarchical cluster-based wireless
sensor networks,” Journal of Wireless Mobile Networks, Ubiquitous Computing, and Dependable Applications (JoWUA), vol. 11,
no. 3, pp. 70–86, Sep. 2020, doi: 10.22667/JOWUA.2020.09.30.070.
[2] V. T. Vu, T. V Quyen, L. H. Truong, A. M. Le, C. V Nguyen, and M. T. Nguyen, “Energy efficient approaches in wireless sensor
networks,” ICSES Transactions on Computer Networks and Communications, vol. 6, no. 1, pp. 1–10, Mar. 2020.
[3] S. Bharany et al., “Energy-efficient clustering scheme for flying ad-hoc networks using an optimized LEACH protocol,”
Energies, vol. 14, no. 19, p. 6016, Sep. 2021, doi: 10.3390/en14196016.
[4] S. El Khediri, R. U. Khan, N. Nasri, and A. Kachouri, “Energy efficient adaptive clustering hierarchy approach for wireless sensor
networks,” International Journal of Electronics, vol. 108, no. 1, pp. 67–86, Jan. 2021, doi: 10.1080/00207217.2020.1756454.
[5] A. M. Jubair et al., “Optimization of clustering in wireless sensor networks: techniques and protocols,” Applied Sciences, vol. 11,
no. 23, p. 11448, Dec. 2021, doi: 10.3390/app112311448.
[6] S. Sharmin, I. Ahmedy, and R. Md Noor, “An energy-efficient data aggregation clustering algorithm for wireless sensor networks
using hybrid PSO,” Energies, vol. 16, no. 5, p. 2487, Mar. 2023, doi: 10.3390/en16052487.
[7] M. R. Reddy, M. L. R. Chandra, P. Venkatramana, and R. Dilli, “Energy-efficient cluster head selection in wireless sensor
networks using an improved grey wolf optimization algorithm,” Computers, vol. 12, no. 2, p. 35, Feb. 2023, doi:
10.3390/computers12020035.
[8] R. Bharathi, S. Kannadhasan, B. Padminidevi, M. S. Maharajan, R. Nagarajan, and M. M. Tonmoy, “Predictive model techniques
with energy efficiency for IoT-based data transmission in wireless sensor networks,” Journal of Sensors, vol. 2022, pp. 1–18,
Dec. 2022, doi: 10.1155/2022/3434646.
[9] S. R. Lahane and K. N. Jariwala, “Secured cross‐layer cross‐domain routing in dense wireless sensor network: a new hybrid based
clustering approach,” International Journal of Intelligent Systems, vol. 36, no. 8, pp. 3789–3812, Aug. 2021, doi:
10.1002/int.22438.
[10] O. Deepa and J. Suguna, “An optimized QoS-based clustering with multipath routing protocol for wireless sensor networks,”
Journal of King Saud University - Computer and Information Sciences, vol. 32, no. 7, pp. 763–774, Sep. 2020, doi:
10.1016/j.jksuci.2017.11.007.
[11] S. Uppalapati, “Energy-efficient heterogeneous optimization routing protocol for wireless sensor network,” Instrumentation,
Mesures, Métrologies, vol. 19, no. 5, pp. 391-397, Oct. 2020, doi: 10.18280/i2m.190510.
[12] P. Maheshwari, A. K. Sharma, and K. Verma, “Energy efficient cluster based routing protocol for WSN using butterfly
optimization algorithm and ant colony optimization,” Ad Hoc Networks, vol. 110, p. 102317, Jan. 2021, doi:
10.1016/j.adhoc.2020.102317.
[13] S. Doostali and S. M. Babamir, “An energy efficient cluster head selection approach for performance improvement in network-
coding-based wireless sensor networks with multiple sinks,” Computer Communications, vol. 164, pp. 188–200, Dec. 2020, doi:
10.1016/j.comcom.2020.10.014.
[14] A. Mehbodniya, S. Bhatia, A. Mashat, M. Elangovan, and S. Sengan, “Proportional fairness based energy efficient routing in
wireless sensor network,” Computer Systems Science and Engineering, vol. 41, no. 3, pp. 1071–1082, Nov. 2022, doi:
10.32604/csse.2022.021529.
[15] S. R. Lahane and K. N. Jariwala, “Secured cross‐layer cross‐domain routing in dense wireless sensor network: a new hybrid-based
clustering approach,” International Journal of Intelligent Systems, vol. 36, no. 8, pp. 3789-3812, 2021, doi: 10.1002/int.22438.
[16] L. Bharathi, N. Sangeethapriya, R. Sasikala, and A. Srinivasan, “Quality aware heterogeneous energy efficient clustering protocol
for wireless sensor network,” International Journal of Conceptions on Electronics and Communication Engineering, vol. 5, no. 1,
May 2022.
[17] P. Durgadevi, T. Veeramakali, and M. S. Glory, “Energy efficient distributed cooperative cluster based communication protocol
in wireless sensor networks,” International Journal of Innovative Technology and Exploring Engineering (IJITEE), vol. 9, no. 6,
pp. 754-758, Apr. 2020, doi:10.35940/ijitee.F3889.049620
[18] N. Malisetti and V. K. Pamula, “Energy aware cluster-based routing in WSN using hybrid pelican-blue monkey optimization
algorithm,” Evolutionary Intelligence, Feb. 2024, doi: 10.1007/s12065-023-00903-6.
[19] N. D. Tan, D.-N. Nguyen, H.-N. Hoang, and T.-T.-H. Le, “EEGT: energy efficient grid-based routing protocol in wireless sensor
networks for IoT applications,” Computers, vol. 12, no. 5, p. 103, May 2023, doi: 10.3390/computers12050103.
[20] A. Ali and M. Vadivel, “Pegasis double cluster head hybrid congestion control in wireless sensor networks,” Journal of
Communications Software and Systems, vol. 17, no. 3, pp. 221-231, Sep. 2021, doi: 10.24138/jcomss-2021-0032.
[21] S. Gopinath, E. V. Boopathy, S. Pragadeswaran, S. Madhumitha, and N. Sureshkumar, “Location based energy efficient routing
protocol for improving network lifetime in WSN,” SSRG International Journal of Electrical and Electronics Engineering, vol. 10,
no. 2, pp. 84-91, Feb. 2023, doi: 10.14445/23488379/IJEEE-V10I2P108.
[22] M. Gurupriya and A. Sumathi, “Dynamic clustering in wireless sensor networks using hybrid jellyfish optimization-leach
protocol,” Dynamic Systems and Applications, vol. 30, no. 11, pp. 1698-1718, Nov. 2021.
[23] R. Dogra, S. Rani, Kavita, J. Shafi, S. Kim, and M. F. Ijaz, “ESEERP: enhanced smart energy efficient routing protocol for
internet of things in wireless sensor nodes,” Sensors, vol. 22, no. 16, p. 6109, 2022, doi: 10.3390/s22166109.

Int J Reconfigurable & Embedded Syst, Vol. 13, No. 3, November 2024: 774-785
Int J Reconfigurable & Embedded Syst ISSN: 2089-4864  785

[24] M. Wu, Z. Li, J. Chen, Q. Min, and T. Lu, “A dual cluster-head energy-efficient routing algorithm based on canopy optimization
and K-means for WSN,” Sensors, vol. 22, no. 24, p. 9731, 2022, doi: 10.3390/s22249731.
[25] S. Pradeep, Y. K. Sharma, C. Verma, S. Dalal, andC. Prasad, “Energy efficient routing protocol in novel schemes for performance
evaluation,” Applied System Innovation, vol. 5, no. 5, p.101, 2022, doi: 10.3390/asi5050101.

BIOGRAPHIES OF AUTHORS

Haritha K. Sivaraman is an assistant professor, in Department of Electronics and

Communication Engineering, RajaRajeswari College of Engineering, Bangalore, India. Her
major interests are wireless sensor networks, energy efficient designs in sensor networks,
internet of things, and wireless communication. She can be contacted at email:
[email protected].

Rangaiah Leburu is a professor in Department of Electronics and

Communication Engineering, RajaRajeswari College of Engineering, Bangalore, India. His
major interests are VLSI, wireless communications, sensor networks, Ad-Hoc networks, and
Routing protocols in wireless networks. He can be contacted at email: [email protected].

Energy-efficient clustering and routing using fuzzy k-medoids … (Haritha K. Sivaraman)