Enhancing the Quality of Service in MANET based

on Optimization based Routing Protocol

NedumaranArappali1 1 1
, Daniel Getachew Tadesse , Abdulkerim Seid Endris , and
* 2
P.Karthika
1
Department of Electrical and Computer Engineering, Wollo University,
Kombolcha Institute of Technology, Ethiopia
1*
maran.van@gmail.com, danielgetachew@kiot.edu.et/
danielgetachew@gmail.com, abdlkerimseid@kiot.edu.et/
kerimsha@gmail.com
2
Adepartment of Computer Applications, Kalasalingam Academy of Research
and Education, Krishnankoil, india
2
karthikasivamr@gmail.com

* Corresponding Author: P.karthika, karthikasivamr@gmail.com

Abstract. The QoS (quality of service) of a mobile adhoc network

(MANET) is determined by a number of factors, the most important of
which is the quality of the links among the routes. The ratio of connection
failure decreases throughput efficiency, which has a direct impact on QoS.
Many industrial units provide a variety of services that can be retrieved
through a MANET. This enables industrial things and users to access
services through several mobile devices. Since the Manet nodes travel at
different speeds and in different directions, the frequency of connection
failure will always be higher. However, a variety of routing protocols have
been addressed previously to support data transmission, but they struggle
when it comes to achieving higher QoS efficiency. This paper presents an
optimization-based route selection algorithm to increase the efficiency of
route selection and maximize QoS. The method measures the likelihood of
connection availability by using the position information, mobility speed,
and direction of numerous nodes. In addition, the value of traffic at each
node, the neighbors of each node, and the frequency of transmission in each
hop are used to calculate the QoS Measure.These parameters are built by
using the k-mean clustering algorithm, the final trust for each node will
calculate, and nodes will classify based on the level of trust. The results
show that the proposed service model has lower communication overhead
and, thus, smaller energy consumption amount is achieved when compared
with recently proposed models. The proposed approach decreases the
energy consumed up to about 50% than the standard AODV protocol.
Keywords: Mobile Adhoc Network, Quality of Service, K-means
Clustering, Optimization, Energy Consumption.
1 Introduction

MANET is a collection of mobile nodes communicating without a static physical

structure. MANETs have a variety of striking features such as topology, quick
setup and wireless multihop communication. MANET is suited for several time-
sensitive applications[1, 2] by all these features. Ad hoc networking provides a
promising connectivity facility that is difficult to construct physical infrastructure.
Moreover, MANETs allow mobile nodes, with no physical structure or
administrative activity, to exchange information. These networks are thus
dynamic, automatically structured and automatically configured, enabling nodes to
switch arbitrarily during communication. In Industry 4.0, too, MANETs play an
important role. These networks can be expanded and combined with emerging
skills such as cloud technology, IoT and engineering techniques in order to build
intelligent industrial automation applications[3, 4]. Figure 1 shows the
arrangement of the MANET. Implementing a stable routing protocol, ensure QoS,
because of its complex topology, is a challenge in the MANET[5–7]. MANET
offers an open communiquésubstructure which enables any multi-faceted node to
join the network easily and contribute in data transmission[8]. This open
communication infrastructure causes MANET safety violations [9].

Fig. 1. MANETStructure

IoT presents various Quality of Service (QoS) requirements which does not exist
in conventional homogeneous wireless and wired networks[10]. A built-in QoS
guarantees are required to provide a reliable end-to-end intelligent system that
meets the requirement of a complete acquisition broadcast interpretation action
loop. Therefore, different network mechanisms and protocols need to be
developed for IoT. QoS is considered as one of the most critical networking issues
that have obtained researchers substantial attention for both wired and wireless
networks [11]. QoS is one of the leading research concerns in the field of IoT
networks. However, the restriction of resources of IoT devices is considered as
one of the main obstacles against developing a reliable QoS handling mechanism
[12]. Interactions and data exchange between IoT devices can provide a reliable
trust model where the QoS approach can be built upon [13].
Different approaches have been proposed to design the trust model most of these
 3

proposed mechanisms mainly provide security for IoT networks [14]. Based on
state of the art, there is a lack of study for build trust model by considering QoS
trust and energy consumption. However, the trust model can be built to focus on
QoS parameters, including throughput, delay, packet loss, and energy
consumption. Scalability is also one of the main problems present in IoT networks
that consist of a high number of devices [15]. The chiefinfluence of this paper is
the design of an energy aware QoS trust model for IoT healthcare devices that
considers energy and QoS in building trust relation and reduce the energy
consumption significantly, when updating the trust values between network
members. Furthermore, this study could be effective for researchers who
interested in developing IoT devices in the healthcare system.

2 Related works

Gu et al. [16] the loT is decomposed into three layers: application layer, a core
layer, and sensor layer, from sides of the network structure of loT. Trust
management for a special purpose for each layer is controlled accordingly, and
these purposes include self-organized, effective routing and multi-service one by
one. The proposed mechanism introduced a high computation overhead in
different layers to build and support the trust model, where IoT limited node
resources can be exhausted. In this case, QoS is also not considered where all data
are handled similarly.
Chen et al. [17] extends this model by adding comprehensive simulation
effectiveness, assess state-of-the-art related work. In light of this, a more advanced
attacker model for analysing the flexibility alongsidethese attacks arranges smart
storage management planning for the ability to limit IoT devices. It is due to
scalability with a comprehensive evaluation, addressing the better route to merge
social parallel metrics to evaluate rates for application performance maximization.
The proposed mechanism targets social networks and services. Also, data priority
was not included in the trust management, where all data are handled similarly.
Asiri and Miri [18] employ distributed probabilistic neural networks (PNNs) to
cluster trustworthy IoT devices from malicious nodes. In addition, it outfits the
problem of the cold start in IoT networks by expecting ratings for newly joined
nodes depending on their characteristics and information collected over time. The
processing task is wholly distributed. Proposed model trying to conquer
limitations of other trust models by training the eventuality neural network and
adapt its weights count on 'nodes' resemblance and the quality of data which is
sent.
Saied et al. [19] the approach clarification aims to manage collaboration in a
nonhomogeneous IoT structure encompass nodes with various resource abilities.
To create an association, support, and practicability of trusted elements in a group
of cooperating services. A single point of failure approach has been used where
4

trust manager failure can result in network trust model failure. Also, higher
communication overhead and long computation are assumed in the bootstrapping
period. QoS is also not considered in building the trust model.
Chen et al. [20] a generalized and unified mechanism was introduced to handle the
issue of trust and reputation by utilizing the development of sensor nodes
community in the Wireless Sensors Network(WSN) of IoT in Cyber-Physical
Systems. Furthermore, the proposed mechanism has been introduced for Cyber-
Physical Systems (CPS) an authentication mechanism is also included. However,
QoS approach is not considered where the primary purpose is to deliver secure
data delivery for all types of data in the same manner where high and low priority
are handled in the same way.
Bao and Chen [21] proposed a scalable trust organisation protocol for IoT. It
emphasizes mainly on the social relationships between nodes. Multiple trust
possessions are considered, with community-interest to account, cooperativeness,
and honesty for social interaction. The proposed mechanism depends on social
relationships where power consumption is not included in trust parameters. All
traffic is handled in the same manner with low and high priority exchanged
similarly.

3 Proposed System

Based on the state of art trust models and control approaches, most of these
approaches are designed for a specific purpose environment. The emphasis is
made more on Security issues about applying a reliable trust model to prevent
attacks. Other than that, approaches introducing higher communication and
processing overhead to building the trust model and exchange trust parameters
reliant on a specific environment are discussed. However, there is an identifiable
gap in the studies for QoS, and different priority traffic handling approaches, that
are not introduced as required metrics in building a trust model. Moreover, energy
consumption needs to be considered in term of trust model designing for IoT
devices as a fundamental factor due to its potentialities of effecting on the
efficiency of these devices.

The goal of energy-saving routing protocols is to find a secure energy route from
the Source Node to the BS for the appropriate period of the network lifetime and
functionality. The network's main purpose is to align energy use (EC) using
various techniques and approaches with the purpose of enlarging the functionality
and NL. This paper introduces SFLA Optimized K-means clustering and AODV
routing. SFLA is used to optimize the K-means procedure with this Hybrid-K-
means SFLA-AODV method. The eight key steps in this hybrid-K-means-SFLA-
AODV method include: 1) IoT Mobile Node Growth, 2) SFLA-based cluster
selection, 3) CH weight-based router estimate, 4) AODV router estimate, 5) CH-
received data and 7) data receiving successfully. Figure 1 below shows the flux
 5

map of the Hybrid-K-means-SFLA-AODV method.

Fig. 2. Hybrid-K-means-SFLA-AODV Procedure

3.1 Clustering based on SFLA with K-means

Initializing the particles, calculating fitness features and ultimately updating the
speed and location in SFLA-based clustering algorithms. SFLA is an optimal
iteration-based procedure and the structure is prepared by modernizing the
generations with a population of random decisions and optimal search. The
particle changes its current position and discovers the best result through
reiteration, combining local and global knowledge in the search phase of SFLA,
based on its own historical and linked information on the neighboring particles. In
each reiteration, the first finest rate is gotten by the qualification function, and the
second most excellent value is obtained by the application of SFLA within the
population. Elements are modified by each element. The best local value is
determined using topological neighbours' particles. After the finest double
standards are discovered, the particle modernizes its speed and position with
following formulas.

V id =WV id +C 1 rand() ( P id−X id ) +C2 Rand () ( P gd −X id ) (1)

X id = X id + V id (2 )
6

The particles position

X id , Pid and P gd is the best position amongaltogether
V
particles is where the particle velocity id is located. Two irregular [0, 1] spectrum
features. And positive parameters are constantly called acceleration coefficients.
The inertia weigh is a restriction which is defined by the user to monitor and
influence the impact of the previous real fragment speed. Enormous inactivity
gage for international research while a smaller latency weight calibrates the quest
area in flux and ebb.

Fig. 3. (a) Fig. 3. (b)

Fig. 3. (a). Unintended sensor employment, 2. (b). CH and GH detection, BS
Installation of the molecules: clusters are made using the K-implies group
calculation for the use of centroids advanced by SFLA when the particles are
randomly transmitted in the underlying advance. The isolation of the streamlined
center with all the hubs in the area is the best aspect of CH. As for the base
bunches are selected, and for several 20xNp molecules, the d best estimates are
accumulated to gain the situation of the molecule. This process is revisited. The d
gbst appreciation is achieved when this process is refreshed with specific cycles
when the molecule's location is compared. The isolation of particles and the bunch
centre, if d est is as little as possible, is calculated as a wellness activity.

3.2 EE clustering-Load balancing using SFLA

SFLA improves the grouping inspired by the characteristics of the ants of nature
and the related area of SFLA, which identifies the problem in transportation
systems for the quickest directions. SFLA is a calculation that streamlines the
creation and operation of fluffy animals. A swarm molecule is a population of
moving items, which can be pulled into better positions through the space of
pursuit. Each winged creature refers to it as a "molecule," flying with a certain
speed and moving to the best location worldwide. SFLA is a global search that has
a strong ability to identify hopeful global results. The meeting ended with
MANET's use of SFLA. Fig.3 provides basic SN clustering techniques. Burden
adjustment aims to amplify the system throughput, limit the time of reaction and
steer overhead. The survey network searched a wide range of sensor hubs to
 7

clarify the adaptability, vitality and lives of SNs.

Fig. 4. Basic Grouping approaches of SN

In the picture 3. The SNs are organized into various cluster networks. The CH is
chosen on the basis of the minimum grade value for each cluster network. The
clusterhead is responsible for assembling data from its neighboring nodes and for
transmitting it to another CH in various networks. The CHs transmit a
confirmation message which contains a time slot plan for their cluster members to
be used during the stage of communication.
Given the adjustments, adjustments and topographic transmission of CH
transmitters and recipients, groups are formed. Each hub knows when the ball can
be transmitted, according to the availability schedule. The CHs collect messages
from all of their groups, supply them and send the results to the BS.
Cluster Head – A head node among the same member of the cluster and other
CHs is called CH.
Cluster Members– In the same cluster networks, cluster members share the same
CH information.

3.3 Shuffled frog-leaping algorithm (SFLA)

Another organically oriented meta heuristic approach was initially industrialised

by Eusuff and Lansey and is based on the second clustering approach. [22]named
SFLA. The SFLA principally developed as a population-based metaheuristic
capable of performing a cognizant heuristic search through any mathematical
function seeking solutions for combinatorial optimization problems. It can be
employed to solve several complicated optimization problems, with the essence of
the non-linearity, non-differentiability and multi-modality [23]. It associates the
advantages of the genetic-based memetic algorithm with the PSO algorithm. In
this algorithm, a virtual population of possible solutions is generated by a set of
frogs (solutions), which segmented to dissimilar subgroups called memeplexes, by
each a local search is performed. The separate frogs hold ideas inside any
memeplex that may be influenced by other frogs' ideas. After a predefined number
of paces of memetic evolution, ideas are transmitted through a process of
memeplexes. The local quest and process of mixing continue until the specified
8

convergence requirements are met. [23]. The SFLA can be briefly described thru
the following steps:

1. Initialization: Select m and n, where m represents the amountof

memeplexes and n denotes the amount of frogs in each memeplex.
Thus, the total sample size, F would be F=m⨯ n.

2. Population generation: An initial population of F frogs is created

random uniformly. For a S- dimensional problem, each frog i is
represented by S variables, such as Fi=(fi 1 , fi 2 ,... , fiS) .
3. Sorting and distribution: The frogs are first classified according to
their fitness values in descending order, after which the entire
population is grouped into memeplex, each with n frogs. In this step,
the first frog is shifted to the first memeplex, the second frog is shifted
to the second, frogm shifts to memeplex.

4. Memeplex evolution: This phase is based on a local quest, in which

the FR and FW are respectively found in each memeplex. The frog is
identified separately as Fg with global best fitness. The Eq. (3) and (4)
are then implemented to increase frog health to the worst possible
exercise.

Di=rand ()×(F g−F w ) (3)

i+1 i
F w =F w + Di ,(Dmin ≤ Di ≤ Dmax ) (4)

Where rand () is random number among zero and 1, Dmin and

Dmax are the minimum and the maximum displacement allowed in a
frog's position respectively. If this process generates a better solution, it
substitutes the worst frog; else, if Eq. (3) and (4) do not increase the
worst solution, Fb of Eq. (3) is replaced with Fg adjusting to Eq. (5):

Di=rand ()×(F g−F w ) (5)

5. Shuffling:The new population is classified according to their fitness

in descending order after a predefining sum of steps to memeplex
evolution all the memeplex frogs are collected.

6. Stop condition:The algorithm stops when a global solution or a

 9

fixed iteration sum has reached. Move to stage (2) otherwise and
repeat.

3.4 AODV-Routing Protocol

The routing protocol would be used in the ad hoc network by mobile nodes. The
AODV is intended to reduce overhead and traffic dissemination. The AODV
Routing Protocol covers two route discovery and route maintenance features. The
Route Discovery role decides on the identification of the fresh route and Route
Maintenance determines the detection of connection breakings and repair of an
established route. The reactive protocol does not keep the route table permanently.
AODV can easily analyze network topology changes.

Fig. 5. AODVData Transfer

4 Results and Discussion

By using the 2017 MATLAB software tool (for simulation purposes), the AODV-
K-means-SFLA device has introduced i5 PC with 8 GB of RAM memory space.In
that proposed method, SFLA used for energy efficient of load balancing and the
routing is carried out by the AODV protocol. The following Table 1
demonstrations the simulation parameters which is used in the AODV-K-means-
SFLA methodology. For knowing the better performance of this AODV-K-means-
SFLA method, it is compared with the only AODV method.
10

Table 1. Simulation factors.

Parameter Value
Initial energy of sensor nodes 0.5 J
Area 250*250 m2
Sensor nodes 200
Sum of simulation iterations Length(msg)
Communication range of each 50 nm
node
ET 0.2 PJ/bit/ m2

ER 0.1 PJ/bit/ m2
Packet size 4000 bits
Message size 200 bits

Alive nodes

Living nodes are called the node which has sufficient energy to transmit the
information (chip text). A more efficient data transmission in MANET needs more
live nodes.

Dead nodes

The nodes with no energy for information transmission are called dead nodes. It
degrades the whole network's efficiency and should reduce dead nodes to prevent
the packet loss.

Energy consumption

The total energy needed for each node to provide the message through the path
extracted in the AODV routing (Cipher text).

Ec =E−( E T −E R ) (3)

Where, MANETenergy consumption is signified as Ec is defiend as the total sum

of energy, the transmitting and getting energy is signified as ET and E R
respectively.

Throughput
 11

The performance is known as the amount of up-and-coming messages sent to the

destination and should be high for an efficient data transmission.
T H =N T × P L (4)

Where, T H is signified as throughput of the MANET, total amount of rounds as

signified as N T and the packet length is signifiedas P L.
Initially, Figure 5-9 shows the graphical results of proposed optimization
technique against AODV-K-means protocol in terms of energy consumption, total
packet send, dead nodes, alive nodes and throughput. The maximum rounds i.e.
5000 rounds are considered in this experimental validation.

Fig. 6. Performance of proposed method in terms of energy consumption

12

Fig. 7. Performance of proposed method in terms of dead nodes

Fig. 8. Performance of proposed method by means of alive nodes

Fig. 9. Performance of proposed method in terms of throughput

From the results itself, it is clearly stated that the proposed protocol along with
SFLA optimization technique achieved better results than AODV protocol. The
energy consumption is high, while implementing AODV-K-means protocol
without optimization. In addition, the dead nodes are less and alive nodes are high
in the proposed AODV-K-means protocol with optimization technique (SFLA).
However, the AODV protocol without optimization has high dead nodes and low
alive nodes, where the throughput is also less in existing AODVprotocol.
 13

5 Conclusion

MANET is a miscellaneous collection of versatile nodes communicating without a

firm physical structure. Because of various remarkable features, such as dynamic
topology, fast setup, and multihop data transmission, MANETs became more
popular. These distinguishing features make MANETs appropriate for numerous
applications such as ecological monitoring and disaster management. In order to
attain the vision of Industry 4.0 MANETs can also be integrated with emerging
skills, such as cloud computing and IoT. A secure and reliable data transmission
that must satisfy a required QoS is required for all MANET sensitive realtime
applications. MANET continuously studies minimizing energy consumption and
protocol communication is identified as one of the main causes of increasing
power consumption. In order to achieve these difficult goals, a routing protocol
needs to be developed to improve the MANET QoS. In this document we have
proposed a K-means clustering optimization based routing protocol with the SFLA
to improve the overall efficiency of the MANET. It takes into account a number of
factors, such as congestion control, packet loss reduction, malicious node
detection, and safe data transmission, in order to intensify the QoS of the
MANET. Simulations in the MATLAB analyze the presentation of the proposed
protocol. Our simulation results demonstrate the superior proposed routing
protocol over the existing approach. In future, we will emphasize security
algorithms through encryption, decryption, and blockchain solutions to ensure the
MANET's high safety.

References

1. M. D. Sirajuddin, C. Rupa, and A. Prasad: Advanced Congestion Control Techniques

for MANET. Advances in Intelligent Systems and Computing, vol. 433, pp. 271–279,
2016.
2. L. Femila and M. MarsalineBeno: Optimizing transmission power and energy efficient
routing protocol in MANETs. Wireless Personal Communications, vol. 106, no. 3, pp.
1041– 1056, 2019.
3. B. Chen, J. Wan, L. Shu, L. Peng, M. Mukherjee, and B. Yin: Smart factory of Industry
4.0: key technologies, application case, and challenges. Institute of Electrical and
Electronics Engineers Access, vol. 6, 2018
4. H. Kathiriya, A. Pandya, V. Dubay, and A. Bavarva: State of art: energy efficient
protocols for self-powered wireless sensor network in IIoT to support industry 4.0. In
Proceedings of the 2020 8th International Conference on Reliability, Infocom
Technologies and Optimization (Trends and Future Directions) (ICRITO), pp. 1311–
1314, Noida, India, June 2020.
5. T. Li, J. Ma, and C. Sun: SRDPV: secure route discovery and privacy-preserving
verification in MANETs. Wireless Net- works, vol. 25, no. 4, pp. 1731–1747, 2019.
6. T. Singh, J. Singh, and S. Sharma: Survey of secure routing protocols in MANET.
International Journal of Mobile Net- work Design and Innovation, vol. 6, no. 3, pp.
142–155, 2016.
14

7. S. Hossain, M. S. Hussain, R. R. Ema, S. Dutta, S. Sarkar, and T. Islam: Detecting

Black hole attack by selecting appropriate routes for authentic message passing using
SHA-3 and Diffie Hellman algorithm in AODV and AOMDV routing protocols in
MANET. In Proceedings of the ICCCNT, pp. 1–7, Kanpur, India, June 2019.
8. V. V. Sarbhukan and L. Ragha: establishing secure routing path using trust to enhance
security in MANET. Wireless Personal Communications, vol. 110, no. 1, pp. 245–255,
2020.
9. H. Moudni, M. Er-rouidi, H. Mouncif, and B. El Hadadi: Secure routing protocols for
mobile ad hoc networks. In Proceedings of the IT4OD, pp. 1–7, Fez, Morocco, March
2016.
10. Safari, E.; and Babakhani, M.: Web service and dynamic pricing competition.
International Journal of Industrial Engineering Computations, 6(1), 99-116.
11. El-Sayed, H.; Mellouk, A.; George, L.; and Zeadally, S.: Quality of service models for
heterogeneous networks: Overview and challenges. Annals of Telecommunications-
Annales des Télécommunications, 63(12), 639-668.
12. Awan, I.; Younas, M.; and Naveed, W: Modelling QoS in IoT applications. Proceeding
of the 17th International Conference on Network- Based Information Systems (NBIS).
Salerno, Italy, 99-105.
13. Guo, J.; Chen, R.; and Tsai, J.J.: A survey of trust computation models for service
management in internet of things systems. Computer Communications, 97(1), 1-14.
14. Al-Shammari, B.K.J.; Al-Aboody, N.; and Al-Raweshidy, H.S.: IoT traffic
management and integration in the QoS supported network. IEEE Internet of Things
Journal, 5(1), 352-370.
15. Alhasan, A.; Audah, L.; Alabbas, A.: Energy overhead evaluation of security trust
model for IoT application. Journal of Theoretical and Applied Information Technology,
98(1), 69-77.
16. Chen, I.R.; Bao, F.; and Guo, J.: Trust-based service management for social internet of
things systems. IEEE Transactions on Dependable and Secure Computing, 13(6), 684-
696, 2015.
17. Chen, I.R.; Guo, J.; and Bao, F.: Trust management for SOA-based IoT and its
application to service composition. IEEE Transactions on Services Computing, 9(3),
482-495, (2014).
18. Asiri, S.; and Miri, A.: An IoT trust and reputation model based on recommender
systems. Proceeding of the 14th Annual Conference on Privacy, Security and Trust
(PST). Auckland, New Zealand, 561-568, (2016)
19. Saied, Y.B.; Olivereau, A.; Zeghlache, D.; and Laurent, M.: Trust management system
design for the internet of things: A context-aware and multi-service approach.
Computers & Security, 39(11), 351-365, (2013).
20. Chen, D.; Chang, G.; Sun, D.; Li, J.; Jia, J.; and Wang, X.: TRM-IoT: A trust
management model based on fuzzy reputation for internet of things. Computer Science
and Information Systems, 8(4), 1207-1228, (2011).
21. Bao, F.; and Chen, I.R.: Trust management for the internet of things and its application
to service composition. Proceeding of the IEEE International Symposium on a World of
Wireless Mobile and Multimedia Networks (WoWMoM). San Francisco, CA, USA, 1-
6, (2012).
22. Eusuff, H., Lansey, M., & Pasha F.: Shuffled frog-leaping algorithm: A memetic meta-
heuristic for discrete optimization. Engineering Optimization, 38(2), 129-154, (2006).
23. Khorsandi, A., Alimardani A., Vahidi B., &. Hosseinian S.H.:. Hybrid shuffled frog
leapingalgorithm and nelder–mead simplex search for optimal reactive power dispatch.
IET GenerationTransmission & Distribution, 5(2), (2011).