DOI: 10.5281/zenodo.

10394448

NODE CLUSTERING TECHNOLOGY FOR ENHANCEMENT THE

QUALITY OF SERVICE OF WIRELESS NETWORK

MOAID ABDULMOHSEN HAMEED ALNAJEM 1, AMEL MEDDEB MAKHLOUF 2

and AHMED FAKHFAKH 3
1, 2, 3
National School of Electronic and Telecommunication of Sfax, Tunisia.
Email: 1 [email protected], 2 [email protected], 3 [email protected]

Abstract
The performance of a mobile ad hoc network (MANET) depends on the volume of traffic and node speed. There
is always room for improvement, especially when routing and node energy consumption are tightly coupled.
Traditional routing techniques are used to increase performance metrics like throughput and latency. This study
links the base station node to hosting nodes with 150, 250, and 350 numbers deploying four cluster heads to
provide AODV-based routing using a clustering strategy. A far-near base approach is used to establish
communication between the nodes. The enhanced clustering method have improved the throughput from 83.92%
(baseline method prior to node clustering) to 92.474 %.
Keywords: Host, AODV, ADHOC, FFNN, Optimization, Clustering.

1. INTRODUCTION
One of the most active research fields in Wireless Sensor Networks (WSNs) is the creation of
low-power electrical devices with wireless communication capabilities. WSNs are networks of
self-contained, widely scattered sensors that are positioned all over a region of interest and
used to accidentally or consciously detect a phenomenon. It took a lot of time and effort to
design, build, and implement these sensor networks because of the special needs for sensing
and monitoring in real-time applications. Wireless modules with a CPU, a transceiver, power,
and memory are built into these nodes. A sensor mode with a range of sensors is installed on
the node depending on the application, such as environmental monitoring [1], surveillance [2],
military applications, transportation automation, health [3], and industrial applications [4]. The
efficient use of stored energy is one of the tightest criteria for these nodes. Various clustering
methods have been used to construct a number of node energy management solutions for WSNs
[5, 6]. In a WSN cluster, each Cluster Head (CH) is in charge of gathering data from the nodes
and sending it to the sink (base station). Sensors are often placed near together to meet coverage
requirements. As a result, certain nodes can enter a sleep state and conserve plenty of energy.
CHs may be chosen at random or in accordance with a set of criteria. The sort of cluster head
being used has a big impact on how long WSNs last. The optimal CH is the one that is closest
to the base station, has the most neighbours, and has the most energy left. MADM (Mosaic
Analysis with Double Markers) techniques [7–10] handle the difficult task of simultaneously
analyzing each of these factors in the selection of CHs. In order to efficiently handle a range of
decision-making difficulties in science, engineering, and social science, many MADM
techniques have been created. Based on a range of standards and criteria, these strategies offer
measurable statistics for possible outcomes. It is well acknowledged that it might be

644 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

challenging to identify all of the criteria's exact values in real-time. Fuzzy-based MADM
techniques in other works [11–13] have shown to be effective and practical in these situations.
In the current work, we tried to use these techniques to make WSNs last longer. Two techniques
are used to enhance the performance of computer networks, particularly adhoc networks:
cooperative routing and clustered routing. However, these methods accelerate packet delivery
between nodes by using the shortest path. Routing protocols like AODV (Ad hoc On-Demand
Distance Vector Routing) are crucial in this endeavor because they reduce route losses and
regulate packet routing. Adhoc networks do not use more advanced routing protocols; instead,
they only use AODV, DSDV, and DSR. However, another concern is how mobility affects the
effectiveness.
1.1 Model Implementation
A clustering strategy model with 150, 250, and 350 nodes is now mobilising each node
individually. The way the model is built allows for the use of routing protocols like AODV. The
model is built using Table 1 as a reference. Data transmission without the aid of clustering
technology was employed when the first senior was established, but when the second senior
was made, data transmission with the help of clustering technology was tested for success.

Figure 1 a: Network Topology Outline Demonstrating the Host Nodes base Station One
Cluster Head in the Group of Randomly Moving Nodes

645 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

Figure a1 represent the network topology which consists of host nodes that supposed to
communicate with each other to transmit the data to the main node (base station) which is
encoloured with dark orange.
While at figure 1b, the clustering is being made to gather the nodes in four clusters where every
cluster is sending the data to the cluster head (encoloured with brown).
Table 1: Model Specifications and Parameters
Object Information
Nodes number 150, 250 and 350
Cases of Clusters 1 and 4
Workspace size 500meter x 500 meter
Routing algorithm AODV
Stop Time 30 seconds
Antenna Type Omni directional antenna
Coverage of node 80 m
Speed of Nodes (10 km/hour)
Movement pattern Random motion
Data Transmission Cooperative routing
According to Table 1, the model is used in practise with a random speed of 10 m/30 sec for
each node. All nodes may access the base station without using any clustering methods (see
Figure 1 a). The nodes will then be connected to the FOUR cluster heads at the base station,
one at the center of each cluster (see Figure 1 b). Host nodes traverse the arena while linked to
the closest cluster head.

Figure 1b: Network Topology Outline Demonstrating the Host Nodes and Cluster
Heads

646 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

Four clusters were made depending on the topography of the system which is rectangular where
in every corner we fixed a cluster to cover the entire arena. The following pseudo code is used
to create the clusters:
Pseudo code
Start program
#determine size of topology areax, y;
x: width , y: height;
#produce nodes with N number = 300;
#fix speed s for node to = 10; sfix the MD motion diraction
= random; fix the NC number of clusters = 4;
N
#Divide the nodes among NC to produce the CN per cluster nodes =
NC
300
= = 75;
4
#fix the cluster heads
CH, cluster head number = 4;
#fix the nodes in cluster for logical mount CN new = CN old − 1;
#Find CR the centriod of the cluster; and fix CH to CR;
#Assignment of node:
for Node ! = CH
distance 1 = distance between Node and CH1
distance 2 = distance between Node and CH2
distance 3 = distance between Node and CH3
distance 4 = distance between Node and CH4

Location = get (minimum (distance1, distance2, distance3, distance4))

if Location = distance1−→ set connection between Node with CH1
else if Location = distance2 → Node with CH2
else if Location = distance3 → Node with CH3
else if Location = distance4 → Node with CH4
End program

2. PERFORMANCE EVALUATION
2.1 Performance Measures
The following measures are used to assess networking performance:
PDR: The packet delivery rate (PDR %) is the proportion of packets that were successfully
transmitted throughout the network, as assessed for each network node. Eq.13 contains details
on packet delivery rate.
𝑁
𝑃𝐷𝑅 = × 100% (13)
𝑇

Where
 T is the total number of packets travelling through the network from source nodes to
destination nodes.
 N represents all of the packets that were received at the destination station.
Packets that are dropped during transmission from the source node to the destination nodes
make up the total number of dropped packets (DP) (Eq.14).

647 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

𝐷𝑃 = 𝑀 − 𝑁 (14)
Where
M represents all of the created packets.
Latency: The time it usually takes for a packet to go from its source to its destination, measured
in seconds after the destination node receives the packet and verifies that it got there.
The number of packets that a source node sends to a destination node is known as throughput
(Eq.15).
𝑁
𝑇ℎ = × 100% (15)
𝑀
PDoR: The packet drop rate is quantity of packets discarded over the connection duration
(transmission interval) (Eq.16).
𝐷𝑃
𝑃𝐷𝑜𝑅 = (16)
𝑇
Results for none cluster and standard cluster methods are illustrated in Table 2 and Table 3
respectively.
The number of transmitted packets is same in all the cases of nodes (all scenarios). (See Figure
2).
4
10
2

1.8

1.6

1.4

1.2
Tx packets

0.8

0.6

0.4

0.2

0
150 250 550
Number of nodes

Figure 2: Number of Transmitted Packets vs the Number of Nodes in all Cases. X Axis
is representing the Cases of Nodes Number Variation, Y Axis is representing the
Number of Packets. No Clustering Technology

648 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

Table 2: No Clustering Topology Performance Metrics

Number of
Sent Received Drops Delivery Rate Drop Rate Delay(ms) Throughput
Nodes
150 2240 1412 828 47.0666667 27.6 13.55 63.0357143
250 8398 7048 1350 234.933333 45 18.58 83.924744
550 19528 15894 3634 529.8 121.133333 33.6 81.3908234
In this scenario, it is absorbed that time delay is increasing when the number of nodes increases,
same for the number of drop packets and packet drop rate. However, throughput is slightly
higher in case of 250 packets and that is due to the random mobility of the nodes.

16000

14000

12000

10000
Rx packets

8000

6000

4000

2000

0
150 250 550
Number of nodes

Figure 3: Number of Nodes vs Number of Received Packets for No Clustering Topology

600

500

400
Delivary Rate

300

200

100

0
150 250 550
Number of nodes

Figure 4: Number of Nodes vs Number of Delivery Ratio of Packets for No Clustering

Topology

649 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

140

120

100
Drop Rate

80

60

40

20

0
150 250 550
Number of nodes

Figure 5: Number of Nodes vs Number of Drop Ratio of Packets for No Clustering

Topology

4000

3500
Number of Dropped Packets

3000

2500

2000

1500

1000

500

0
150 250 550
Number of nodes

Figure 6: Number of Nodes vs Number of Dropped Packets for No Clustering Topology

650 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

35

30

25
Time Delay (ms)

20

15

10

0
150 250 550
Number of nodes

Figure 7: Number of Nodes Latency for No Clustering Topology

90

80

70

60
Throughput

50

40

30

20

10

0
150 250 550
Number of nodes

Figure 8: Number of Nodes Throughput for No Clustering Topology

651 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

2.3 Clustering Technique

Table 3: Standard Clustering Topology Performance Metrics
Number of
Sent Received Drops Delivery Rate Drop Rate Delay (s) Throughput
Packets
150 2240 1744 496 58.1333333 16.5333333 12.22 77.8571429
250 8398 7766 632 258.866667 21.0666667 17.84 92.4743987
550 19528 16010 3518 533.666667 117.266667 43.37 81.9848423
In this scenario, it is absorbed that all results is being enhanced after introduction of the
clustering method over the standard method of the previous scenario.

18000

16000

14000

12000
Rx packets

10000

8000

6000

4000

2000

0
150 250 550
Number of nodes

Figure 9: Number of Nodes vs Number of Received Packets for Standard Clustering

Topology

652 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

4000

3500
Number of Dropped Packets

3000

2500

2000

1500

1000

500

0
150 250 550
Number of nodes

Figure 10: Number of Nodes vs Number of Dropped Packets for Standard Clustering
Topology

600

500

400
Delivary Rate

300

200

100

0
150 250 550
Number of nodes

Figure 11: Number of Nodes vs Number of Delivery Ratio of packets for Standard
Clustering Topology

653 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

120

100

80
Drop Rate

60

40

20

0
150 250 550
Number of nodes

Figure 12: Number of Nodes vs Number of Drop Ratio of Packets for Standard
Clustering Topology

45

40

35

30
Time Delay (ms)

25

20

15

10

0
150 250 550
Number of nodes

Figure 13: Number of Nodes Latency for Standard Clustering Topology

654 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

100

90

80

70
Throughput

60

50

40

30

20

10

0
150 250 550
Number of nodes

Figure 14: Number of Nodes Throughput for Standard Clustering Topology

3. CONCLUSION
Wireless communication over short distances is made possible by mobile ad hoc networks that
link the hosts. The amount of data being transmitted and the number of participating network
nodes both have an impact on communication quality. In this study, the efficacy of cluster
routing with and without the use of two seniors is examined. At 150, 250, and 350 nodes,
respectively, the network performance of the routing protocol using none clustering is assessed.
The base station node, which is situated in the center of the network coverage region, is
receiving data from host nodes. If nodes are inside the base station's communication range,
they can communicate with it. If the node is beyond the coverage region, multi-hop connections
are established through other valid nodes. Contrarily, when four clusters are established for the
same number of nodes, cluster-based routing is employed. The near-far theorem is used to
reconstruct the link between the host nodes and cluster head nodes, with a 10 km/h speed limit
for each node. Thanks to PSO-FFNN, the mobile node selects the cluster head more rapidly
and effectively. The findings demonstrate that as compared to no clustering routing, cluster-
based routing has improved network performance. The effectiveness of the clustering strategy
has greatly increased as a result of the improved clustering technique used in this study
(algorithm as shown the pseudo code). Utilizing a clustering strategy improves efficiency by
0.72%.

References
1) R. Rajendran An optimal strategy to countermeasure the impersonation attack in wireless mesh network Int.
J. Inf. Technol. (2021), pp. 1-6
2) R. Sivakami, S.C. Raghava Strong security scheme for single node and colluding nodes byzantine Attacks
in MANETS Ann. Romanian Society for Cell Biol. (2021), pp. 1657-1666

655 | V 1 8 . I 1 2
 DOI: 10.5281/zenodo.10394448

3) D.F. Sittig, H. Singh A socio-technical approach to preventing, mitigating, and recovering from ransomware
attacks Appl. Clinical Informatics, 7 (2) (2016), p. 624
4) M. Raya, J.P. Hubaux Securing vehicular ad hoc networks J. Computer Security, 15 (1) (2007), pp. 39-68
5) Y.C. Hu, A. Perrig, D.B. Johnson Ariadne: a secure on-demand routing protocol for ad hoc networks Wireless
networks, 11 (1) (2005), pp. 21-38
6) K. Sharshembiev, S.M. Yoo, E. Elmahdi Protocol misbehavior detection framework using machine learning
classification in vehicular Ad Hoc networks Wireless Networks, 27 (3) (2021), pp. 2103-2118
7) T. Sushma A review of the cluster based mobile adhoc network intrusion detection system Turkish J. Comput.
Mathematics Education (TURCOMAT), 12 (2) (2021), pp. 2070-2076
8) A. Tami, S. Boukli Hacene, M. Ali Cherif Detection and prevention of blackhole attack in the AOMDV
routing protocol J. Commun. Software Sys., 17 (1) (2021), pp. 1-12
9) N.K. Trivedi, A. Kumar, A. Anand, S. Maheshwari Cross-layer intrusion detection in mobile ad hoc
networks–a survey Ann. Romanian Society for Cell Biol. (2021), pp. 09-20
10) M.S. Usha, K.C. Ravishankar Implementation of trust-based novel approach for security enhancements in
MANETs SN Computer Sci., 2 (4) (2021), pp. 1-7
11) M. Premkumar, T.V.P. Sundararajan Defense countermeasures for DoS attacks in WSNs using deep radial
basis networks Wireless Personal Commun. (2021), pp. 1-16
12) R. Ahmed, Y. Chen, B. Hassan, L. Du CR-IoTNet: machine learning based joint spectrum sensing and
allocation for cognitive radio enabled IoT cellular networks Ad Hoc Networks, 112 (2021), Article 102390
13) E. Anceschi, G. Bonifazi, M.C. De Donato, E. Corradini, D. Ursino, L. Virgili SaveMeNow. AI: a machine
learning based wearable device for fall detection in a workplace Enabling AI Applications in Data Science,
Springer, Cham (2021), pp. 493-514
14) J. Ramkumar, R. Vadivel Multi-adaptive routing protocol for internet of things based ad-hoc networks
Wireless Personal Commun. (2021), pp. 1-23
15) K.E. Lee, J.G. Park, S.J. Yoo Intelligent cognitive radio ad-hoc network: planning, learning and dynamic
configuration Electronics, 10 (3) (2021), p. 254
16) A. Alsarhan, M. Alauthman, E.A. Alshdaifat, A.R. Al-Ghuwairi, A. Al-Dubai Machine learning-driven
optimization for SVM-based intrusion detection system in vehicular ad hoc networks J. Ambient Intell.
Humanized Comput. (2021), pp. 1-10
17) A. Khraisat, A. Alazab A critical review of intrusion detection systems in the internet of things: techniques,
deployment strategy, validation strategy, attacks, public datasets and challenges Cybersecurity, 4 (1) (2021),
pp. 1-27
18) C. Pham, M. Ehsan Dense deployment of lora networks: expectations and limits of channel activity detection
and capture effect for radio channel access Sensors, 21 (3) (2021), p. 825
19) Z.A. Younis, A.M. Abdulazeez, S.R. Zeebaree, R.R. Zebari, D.Q. Zeebaree Mobile ad hoc network in disaster
area network scenario: a review on routing protocols Int. J. Online Biomed. Eng., 17 (3) (2021)
20) I. Khelafa, A. Ballouk, A. Baghdad Control algorithm for the urban traffic using a realtime simulation Int. J.
Electrical and Computer Eng. (IJECE), 11 (5) (2021), pp. 3934-3942
21) P. Wang, Z. Sun, M. C. Vuran, M. A. Al-Rodhaan, A. M. Al-Dhelaan, and I. F. Akyildiz, “On network
connectivity of wireless sensor networks for sandstorm monitoring,” Computer Networks, vol. 55, no. 5, pp.
1150–1157, 2011.

656 | V 1 8 . I 1 2