Proceedings of the International Conference on Smart Electronics and Communication (ICOSEC 2020)

IEEE Xplore Part Number: CFP20V90-ART; ISBN: 978-1-7281-5461-9

A novel supervised learning based neighbor

discovery in MANET
Mobility Prediction in MANET

Saba Farheen N S Dr. Anuj Jain Dr. Virendra Kumar Sharma

Department of Electronics and Department of Electronics and Department of Electronics and
Communication Engineering Electrical Engineering Communication Engineering
Bhagwant University Lovely Professional University Bhagwant University
Ajmer, Rajasthan, India Phagwara, Punjab, India Ajmer, Rajasthan, India
Saba083@gmail.co m a1978jain@gmail.co m virenkrec@gmail.com

Abstract—Mobile ad hoc networks (MANETs) is a multi-hop established and the data packets reach the destination.
wireless network having the capability to self- configure and self- Hence, it is also called as On-demand routing.
organize when affected by node mobility. Routing becomes a
problem in MANET because nodes continuously change their iii. Hybrid Protocols: It is a mix of both proactive and
position due to movement. As a result, a packet path must be reactive routing protocols. The choice of one or the
dynamically adapted according to the node movement. Most other routing depends upon the particular type of
routing protocols in MANET are characterized by low packet application or typical cases.
success ratio and low throughput. This work proposes an
effective routing algorithm assisted by artificial neural networks
Routing becomes a problem in MANET because nodes
for mobility prediction in order to improve the routing continuously change their position due to movement. As a
performances, thereby increasing the packet success ratio and result, the packet path must be dynamically adapted according
throughput. to the node movement. Most routing protocols in MANET are
characterized by low packet success ratio and high
Keywords—Routing; Mobile Ad Hoc Network (MANET); communication overhead. Hence, a sophisticated routing
Mobile Host’s(MH’s); Mobility prediction; Supervised learning protocol is proposed to address this issue.
A most likely application of MANET’s is in the field of
I. INT RODUCT ION emergency communication required during disaster
A Mobile Ad-hoc network is an assembly of the number of management. In the case of earthquakes or flooding, the
nodes that are free to move in any direction. Every single node telecom based infrastructure networks would fail and in this
in the network is independent of the topology and hence can be case, MANET comes to their rescue. So it is really necessary
referred to as an autonomous node. As the nodes are mobile in and helpful to propose a mechanism to address the routing
nature, the network topology changes dynamically, thereby problems in MANET.
each node serves as a router to facilitate packet routing. The
node from which the data packets are generated is called the II. RELAT ED W ORK
source node. The node to which the packets have to be
delivered in the network is called the destination node. Multi- MANET is an infrastructure-less temporary system, shaped
hop routing is employed to facilitate packet delivery amongst by an arrangement of mobile hosts that has no focal
the nodes. Due to node mobility, the routing path for the organization and builds up their particular network
packets changes frequently. This results in a reduction of progressively. In addition, the system's condition has a few
packet delivery ratio. highlights that include additional complexities, for example,
the regular changes in the topology caused by mobile hubs, and
Many routing algorithms have been suggested for MANET also the inconsistency, asset requirement and the data
which is classified into the following transmission constraint of remote channels. Various routing
i. Proactive routing: A route table about the position of protocols and mobility prediction techniques were proposed to
each and every node is built frequently and routing is optimize the network performance for effective steering in
based on it. Due to this, it is also called as table-driven MANET which is as discussed below.
routing. The authors in [1] clarify about the fast propagation of
lightweight remote gadgets, for example, PCs, remote phones,
ii. Reactive Routing: Route establishment is done on
demand by flooding the network with route request and remote sensors, the potential and significance of nomadic
packets. Based on the response, the next node is computing especially mobile Adhoc networking. Owing to the
identified and this process goes on until a fixed path is vigorous topology of Adhoc Network, this paper fundamentally
emphases on numerous directing strategies which are the most

978-1-7281-5461-9/20/$31.00 ©2020 IEEE 716

Authorized licensed use limited to: Newcastle University. Downloaded on October 18,2020 at 15:38:30 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Smart Electronics and Communication (ICOSEC 2020)
IEEE Xplore Part Number: CFP20V90-ART; ISBN: 978-1-7281-5461-9

difficult issue in the present situation. There are many training procedure used is the back-propagation algorithm. As
protocols proposed for MANETs which makes it very hard to the nodes are dynamic in the network, there is no particular
figure out which convention is appropriate for various system path amongst source and the destination node. Hence, mobility
conditions. This paper effectively gives a review of various prediction provides more stable links for communication in
routing protocols. MANETs.
In [2], the authors explain that wireless Sensor Network In [8], the authors explain the importance of protocol
helps in checking different sorts of conditions by detecting the selection and routing to obtain an efficient network.
physical phenomenon. The paper’s real focus lies in expanding Comparisons are made between different Adhoc routing
the network lifetime so that the battery shall not be drained protocols, specifically between AODV and DSDV to have
soon. Clustering the sensor hubs is a powerful procedure for complete statistics about the performance analysis and results.
accomplishing this objective. This paper presents an energy- The different performance parameters used for comparison
productive clustering calculation for a sensor network in light were an end-to-end delay, jitter and throughput. Results show
of the LEACH convention. LEACH is one of the well-known that the effort required for routing with mobile nodes is
cluster-based structures, which has been generally proposed in comparatively less using AODV. Thus, AODV helps in
remote sensor networks. The proposed protocol is recreated reducing network overhead.
and the outcome demonstrates a critical decrease in In [9], the authors explain about multipath routing protocol
organization energy utilization when contrasted with LEACH. based on the stable backbone, focussing on finding the steady
In [3], the author sets out a proposition for DREAM_OLSR way to exchange the data packets. This ,, was carried out in 2
protocol. The convention had been proposed to impact the stages, the first stage involved the determination of correct hub
existing OLSR convention. The convention builds up an having higher enduring bandwidth, remaining energy, link
augmented solution, wherein Distance routing effective quality and dynamicity. Different routing routes were set up
algorithm is employed for optimization thus resulting in utilizing these hubs. Secondly, the backbone development to
increased efficacy and reduced overheads for MANETs. facilitate routing from source to destination was done. If any
In [4], the author presents an enhanced AODV convention. hub in a course tends to fail, at that point, the transmission
In this convention, a stability factor is utilized to monitor the would occur through the substitute way.
rest of the energy of each hub in the setup. When an RREQ In [10], the authors explain about AODV with reliable
message is sent by the source hub it checks its stability factor delivery (AODV-RD). An intelligent routing algorithm is built
during the time it ranges to its neighbouring hub. The AODV called AODV-RD which has an improved packet delivery
directing convention for MANETs that viably diminishes the ratio. Simulation results were used to demonstrate that AODV-
energy utilization and the end defer amid the course revelation RD essentially increases the packet success ratio when
process is suggested. By utilizing these strategies in directing compared to AODV-BR.
convention it is conceivable to diminish the energy and end to While dealing with MANETs, routing becomes a very huge
end defer by around 7% to 24%. problem since the nodes are continuously in movement. Most
In [5], the author explains about one of the most promising of the previously discussed solutions have a lot of unresolved
routing protocols for MANETS named DSR (Distance State issues such as low packet success ratio and high n etwork
Routing). Continuous Hop field Neural Network is utilised for communication overhead. Also, the existing procedures are not
route optimization which can improve usability as well as scalable for large networks since one particular approach does
survivability of MANET. not apply to all kinds of mobility. Although the nodes are
In [6], the author explains about Mobility prediction which random, there is a specific movement pattern underlying the
in case of versatile specially appointed systems is a movement of every node. If a machine learning classifier can
straightforward and powerful plan to foresee when the client be trained to learn the movement pattern and predict the
will leave from the present area and how will it move to the position of the node in the next point of time, this information
next future position. The versatility expectation is can be used for routing. Thus an effective solution using
exceptionally critical for homogeneous and heterogeneous artificial neural networks would be proposed to solve the
systems, dictated by applications in portability administration, routing problems thereby enhancing the routing performances,
smooth handoffs, call confirmation control, and asset booking by increasing the bandwidth, packet success ratio and
for enhanced nature of administration. One approach to address throughput.
the reality of hub versatility is contriving proactive
arrangements that make utilization of portability forecast III. CLUST ERING
calculations to evaluate the development of the hubs. This The process of grouping a given number of nodes within
paper display the most noteworthy vers atility expectation the entire network is referred as clustering, wherein the entire
procedures for MANETs concentrating on their significant plan network gets clustered frequently or after a fixed period of
standards and portrayal. time-based on the proximity of mobile nodes which are
In [7], the author uses neural network technology to foretell dynamic. In a cluster, any node which remains relatively stable
the forthcoming position of the mobile nodes in Ad hoc to all the other nodes would be chosen as a Cluster head.
networks. The proposed method consists of a multi-layered
feed-forward architecture for neighbourhood prediction. Neural

978-1-7281-5461-9/20/$31.00 ©2020 IEEE 717

Authorized licensed use limited to: Newcastle University. Downloaded on October 18,2020 at 15:38:30 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Smart Electronics and Communication (ICOSEC 2020)
IEEE Xplore Part Number: CFP20V90-ART; ISBN: 978-1-7281-5461-9

Fig. 1. Structure of a clustered ad-hoc network [2],[14],[17]

The cluster heads act as a relay nodes for inter-cluster

communication wherein at least two cluster heads are involved
for sending the data packets from one cluster to another cluster
in the entire network. For intracluster communication only one
cluster head is involved which in turn is responsible for
directing the data packets from the source node to the
destination node within the cluster [16],[20]. The structure of a
clustered MANET is as shown in Fig. 1 which differentiates
amongst the cluster heads and the rest of the mobile nodes
within the cluster in the entire network.
The algorithm used for clustering is LEACH which stands
for Low Energy Adaptive Clustering Hierarchy [21], being one
of the energy-efficient algorithms used for clustering. The
algorithm randomly selects the cluster head in each and every
iteration. All the nodes other than the cluster head selects a
nearest cluster head node and seams to that cluster.
In LEACH, cluster head selection is based on the equation
(1).

­ U
° if iH
° § q mod ·
T (i) ®1  U * ¨¨ ¸¸ (1)
° © U ¹
°¯ 0otherwise
A candidate node in the network which desires to be a
Where, cluster head broadcasts a random value ‘q’. This random value
U – Preferred percentage of cluster heads within the ‘q’ would be matched with the calculated threshold value
entire network. essential for a node to be a cluster head. If this random value of
q – Current iteration or the current round. a node is lesser than the threshold value, then the candidate
H – Assembly of nodes that have not been the cluster node itself becomes the cluster head else the candidate node
heads in the preceding iteration. needs to wait for its turn until all the remaining nodes in the
For a node to be the cluster head a value between 0 and 1 network have turned out to be the cluster head at least once.
needs to be chosen by the node. If the chosen value is lesser Once the cluster head is chosen it is given an identification
than the threshold value T(i), then the node turns out to be the number. Using this identification number, the cluster head
cluster head for the present round. A node that has become a sends an advertisement to the remaining nodes so that all the
cluster head in the preceding rounds does not turn out to be a nodes contained by the cluster links to the cluster head. Thus,
cluster head in the subsequent round until all the existing nodes the nodes submit themselves to the cluster head by accepting
in the entire network have become cluster heads at least once. the request to join the cluster head.
The entire process for cluster head selection takes place in During the steady phase, the random value assigned to the
two steps namely setup phase and steady phase. The algorithm cluster head would be compared with the threshold value. As
for these two phases is as follows long as the random value is less er than the threshold value, the
given node is maintained as a cluster head in the network. As
the nodes are mobile leading to a dynamic topology thereby the
entire process of cluster head selection has to be carried out
again.

978-1-7281-5461-9/20/$31.00 ©2020 IEEE 718

Authorized licensed use limited to: Newcastle University. Downloaded on October 18,2020 at 15:38:30 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Smart Electronics and Communication (ICOSEC 2020)
IEEE Xplore Part Number: CFP20V90-ART; ISBN: 978-1-7281-5461-9

IV. PROPOSED M ET HODOLOGY TABLE I. NEURAL NETWORK CONFIGURATION

A machine learning approach in conjugation with clustering Type, Layer M ulti-Layer Perceptron; 3
is employed on the MANET to facilitate efficient packet No of input neurons 4
delivery, which is carried out by predicting the node position to
avoid link breakage and to optimize the node energy. The No of output neurons 1
entire work is split into modules and each module is executed No of hidden Layer 1
step by step one after the other. The flow process for all the
modules is as shown in Fig. 2. No of hidden Layer neurons 13
Learning algorithm LM S
Error rate 0.001
Learning rate 0.9

The performance parameters of the ANN applied network

are compared with a simple MANET with no such
sophistication such as clustering and mobility prediction.
Simulation results would illustrate that the performance
parameters of the ANN applied network would be improvised
effectively.

V. SIMULAT ION RESULT S

Network Simulator v.2 is the platform or the software used
for simulating. The programming must be done in TCL – tool
command language. The snapshots are the results that were
Fig. 2. Flow Process
achieved after step by step execution of all the modules of the
The first step in the workflow is to set up a network based system.
on clustering and selection of cluster head. After setting up the Simulation configuration file as shown in Fig. 3 configures
mobile nodes, position prediction of each and every node is all the seven layers of the OSI reference model. The number of
done by making use of a neural network-based machine nodes is set to the required amount, 802.11 wireless protocol is
learning classifier. Based on the records of the movement of used, a routing protocol is set as AODV and the graphical
each and every node, a sequence of data is learnt and used for region is set to be 900*900 pixels.
predicting the position of the node at a particular instant of time
when the network tries to direct the information packets from
source to destination.
A feed-forward neural network involving at least three
layers of nodes is employed, it makes use of a supervised
learning approach called backpropagation to carry out the
training. Each fragment of data gets processed and based on the
accuracy of the output in comparison with the expected value,
learning happens in the perceptron wherein the connection
weights would get changed accordingly. In fact, this neural
architecture employs N-step prediction for estimating N
annotations of time series. The prediction step is represented by
‘m’, for every ‘m’ the network output needs to be computed.
When m ≠ N, network output value is given as sˆ (t + m) which
would be stored in the input layer. As a result, the system
would be capable of estimating the subsequent value sˆ (t + (m
+ 1)), which corresponds to the prediction step (m+1).
Corrections which results in the least possible error in the Fig. 3. Simulation configuration file in tcl
entire output were used to set node weights.
Table. I show the neural network configuration that is
required for the effective performance of the proposed
algorithm. The ANN applied network is then subjected to
routing either DSDV or AODV based on whether there is intra-
cluster or inter-cluster communication respectively.

978-1-7281-5461-9/20/$31.00 ©2020 IEEE 719

Authorized licensed use limited to: Newcastle University. Downloaded on October 18,2020 at 15:38:30 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Smart Electronics and Communication (ICOSEC 2020)
IEEE Xplore Part Number: CFP20V90-ART; ISBN: 978-1-7281-5461-9

Fig.6 shows the reactive routing instance. Routing happens

based on the proposed method i.e. reactive routing for inter
clusters wherein the path will be, the source nodeÆCluster
head of source nodeÆcluster head of the destination
nodeÆdestination node.

Fig. 4. User input instance

Fig.4 shows the user input instance. After clustering, the

user is asked to enter the source node and the destination node
to which either proactive or reactive routing protocol would be
selected and routing would be carried out.
Fig. 7. Proactive routing instance

Fig.7 shows the proactive routing instance. Routing

happens based on the proposed method i.e proactive routing for
intra cluster’s wherein the path will be, source node to
destination node directly as the route will be already available
in the routing table of the respective cluster head. The obtained
simulation consists of various calculated parameters such as
bandwidth, packet success ratio and throughput.
The plots for all the above-mentioned performance
parameters were obtained. These plots illustrated that the
network performance improves after incorporating mobility
prediction as compared to conventional AODV routing
protocol.
Fig. 5. Network topology after cluster head selection and Clustering [2]
Fig.8 shows that the packet success ratio in the proposed
Fig.5 depicts cluster head selection and clustering where method is better than AODV for a fixed number of nodes. The
packet success ratio increased by 20% after incorporating
cluster heads are selected based on LEACH algorithm [2] and
all other nodes join the respective cluster heads. mobility prediction.

Fig. 6. Reactive routing instance

Fig. 8. Packet success ratio plot

978-1-7281-5461-9/20/$31.00 ©2020 IEEE 720

Authorized licensed use limited to: Newcastle University. Downloaded on October 18,2020 at 15:38:30 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Smart Electronics and Communication (ICOSEC 2020)
IEEE Xplore Part Number: CFP20V90-ART; ISBN: 978-1-7281-5461-9

Table II presents a comparative result analysis of several

parameters such as bandwidth, throughput & packet success
ratio of the network with & without mobility prediction.

VI. CONCLUSION
Due to the use of Artificial Neural Networks for position
prediction and use of predicted position to do routing, the
performance of the routing protocol had increased
significantly. The proposed solution for routing in MANETs is
scalable for relatively massive networks. By using more
complex neural network algorithms for neighbor prediction, the
network can be made fault-tolerant since the probability of a
data packet reaching the destination node is higher, even if one
of the links breaks due to congestion or any other abrupt
activity in the network. These type of highly efficient networks
with no fixed infrastructure have promising application in the
Fig. 9. T hroughput plot field of disaster management and emergency communication.

Fig.9 illustrates that the throughput in the proposed method Ad-hoc network uses wireless links for the data
transmission of packets which subsequently makes it highly
is better than AODV for a fixed number of nodes. Throughput
increased approximately by 25% after incorporating mobility susceptible to attack. Even though more or less some work has
prediction. been done on MANET's security issues, still many issues and
challenges needs are addressed. As it is quite challenging to
offer security due to the absence of centralized architecture or
authority, as key management turns into a problematic
situation. Therefore, special efforts must be made to ensure
both secure and reliable communication.

REFERENCES

[1] Mahima chitkara , Mohammed waseem ahmed, “ Review on MANET :

characteristics, challenges, imperatives and routing protocols”,
International journal of computer science and mobile computing, vol 3,
issue 2, February 2014, pp 199-206.
[2] Monika, Sneha Chauhan, Nishi Yadav, “ LEACH-I Algorithm for
WSN”, International Journal of Innovative Research in Computer and
Communication Engineering ,Vol. 4, Issue 3, March 2016, pp 435 -439.
[3] Gurpreet Singh Saini , Ashish Kots , Manoj Kumar, “ DREAM_OLSR
PROT OCOL (Distance Routing Effective Algorithm for Mobility -
Optimized Link State Routing)”, International Journal of Computer
Fig. 10. Bandwidth plot T rends and T echnology (IJCTT), Volume 4 Issue 5, May 2013, pp 562 -
570.
Fig. 10 shows the bandwidth plot. The bandwidth is [4] Sangeeta Kurundkar, Apoorva Maidamwar, “ An Im proved AODV
Routing Protocol For Mobile Ad-Hoc Networks”, International Journal
increased by 27% after incorporating mobility prediction. As a of Advanced Research in Electrical,Electronics and Instrumentation
result, the network is now more scalable since signals of higher Engineering, Vol. 2, Issue 7, July 2013, pp 270 -278.
frequencies can be used. [5] Hua Yang, Zhimei Li, “ A Routing Optimization T echnology based on
Neural Networks in MANET .” Gullin university of Aerospace
technology, Gullin, China. IEEE journals, 2017, pp 412 -415.
TABLE II. RESULT ANALYSIS
[6] Amit savyanavar, Ram Kale, “ Mobility Prediction: A review of
techniques used in MANET s.” International Journal of Research in
Parame te rs Without With Computer and Communication T echnology, Vol 3, Issue 1, January
Mobility Mobility 2014, pp 332-337.
Pre diction Prediction
[7] Nermin Makhlouf ,” Exploiting Neural Networks for Mobility
Bandwidth(KBps) 3.72 4.69 Prediction in Mobile Ad-Hoc Networks.” Faculty of Electrical and
communication, Brno University of T echnology. Elect rorevue Vol 18,
No 3, June 2016, pp 132-136.
Throughput(KBps) 7.4 9.2
[8] Neha Aggarwal, T eglovy Singh Chohan, Karamveer Singh,Rajan Vohra,
Dr.Shalini Bahel,” Relative Analysis of AODV & DSDV Routing
Packet success 80 100 Protocols for MANET based on NS2”. International Conference on
ratio (%) Electrical, Electronics and Optimization T echniques (ICEEOT ), IEEE
paper, 2016, pp 512-515.
[9] Saba Farheen N S, Anuj Jain, “ Improved Routing in MANET with
Optimized Multi path routing fine tuned with Hybrid modeling”, Journal

978-1-7281-5461-9/20/$31.00 ©2020 IEEE 721

Authorized licensed use limited to: Newcastle University. Downloaded on October 18,2020 at 15:38:30 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Smart Electronics and Communication (ICOSEC 2020)
IEEE Xplore Part Number: CFP20V90-ART; ISBN: 978-1-7281-5461-9

of King Saud University- Computer and information Sciences(2009),

doi: https://doi.org/10.1016/j.jksuci.2020.01.001 .
[10] LIU Jian, LI Fang-min “An Improvement of AODV Protocol Based on
Reliable Delivery in Mobile Ad hoc Networks” Fifth International
Conference on Information Assurance and Security, 2009, DOI
10.1109/IAS.2009.78, pp 154-159.
[11] Deepak Dembla, Dr. YogeshChaba,”Modeling and Analysis of an
Intelligent AODV Routing Protocol based on Route Request
Retransmission Strategy in MANET s”,International Journal of
Computer Applications (0975 – 8887)Volume 30– No.11, September
2011, pp 205-208.
[12] Duan, Shukai, "Small-world Hopfield neural networks with weight
salience priority and memristor synapses for digit recognition." Neural
Computing and Applications 27.4 (2016), pp 837 -844.
[13] Perkins, Charles E., and Pravin Bhagwat. "Highly dynamic destination
sequenced distance-vector routing (DSDV) for mobile computers."
ACM SIGCOMM computer communication review. Vol. 24. No. 4.
ACM, 2004, pp 132-136.
[14] Farooq Aftab, Zhongshan Zhangand Adeel Ahmad. “ Self-Organization
Based Clustering in MANET s Using Zone Based Group Mobility,”
IEEE transactions, VOLUME 5, 2017, pp 563 -571.
[15] Usha R, Premananda B.S and Viswavardhan Reddy K, “Performance
Analysis of MANET Routing Protocols for Military Applications”,
International Conference on Intelligent Computing and Control Systems
ICICCS, 2017, pp 621-627.
[16] Wei Liu, Hiroki Nishiyama, Nirwan Ansari, Jie Yang, and Nei Kato,
“Cluster-Based Certificate Revocation with Vindication Capability for
Mobile Ad Hoc Networks”, IEEE transactions on parallel & distributed
computing”, Vol. 24, 2015, pp 256-259 .
[17] Zhigang Wang, Lichuan Liu, MengChu Zhou, Nirwan Ansari, “ A
Position-Based Clustering T echnique for Ad Hoc Intervehicle
Communication”, IEEE transactions on systems, man, and cybernetics- -
part c:applications and reviews,vol.38, March 2008, pp 126 -132.
[18] Hopfield, John J. “ Neural networks and physical systems with emergent
collective computational abilities.”, Proceedings of the national academy
of sciences, 1982, pp 2554-2558.
[19] D. Gavalas, G. Pantziou, C. Konstantopoulos, and B. Mamalis, “ Lowest
ID with adaptive ID reassignment: A novel mobile ad-hoc networks
clustering algorithm,'' in Proc. 1st Int. Symp. Wireless Pervasive
Comput.,Jan. 2006, pp 5-8.
[20] F. D. T olba, D. Magoni, and P. Lorenz, “Connectivity, energy and
mobility driven clustering algorithm for mobile ad hoc networks,” in
Proc. IEEE GLOBECOM, Nov. 2007, pp 2786-2790.
[21] M. Cai, L. Rui, D. Liu, H. Huang, and X. Qiu, “Group mobility based
clustering algorithm for mobile ad hoc networks,” in Proc. APNOMS,
Aug. 2015, pp 340-343.
[22] M. Ni, Z. Zhong, and D. Zhao, “ MPBC: A mobility prediction -based
clustering scheme for ad hoc networks,” IEEE T rans. Veh. T echnol.,
vol. 60, no. 9, Nov. 2011, pp. 4549-4559.
[23] C.-W. Wu, T .-C. Chiang, and L.-C. Fu, “ An ant colony
optimizationalgorithm for multi-objective clustering in mobile ad hoc
networks,” in Proc. IEEE Congr. Evol. Comput., Jul. 2014, pp. 2963-
2968.
[24] Z. Zhang, K. Long, J. Wang, and F. Dressler, “ On swarm intelligence
inspired self-organized networking: Its bionic mechanisms, designing
principles and optimization approaches,” IEEE Commun. Surveys Tuts.,
vol. 16, no. 1, 2014, pp. 513-537,.
[25] D. T ian, J. Zhou, Z. Sheng, M. Chen, Q. Ni, and V. C. M. Leung,
“Selforganized relay selection for cooperative transmission in vehicular
ad-hoc networks,” IEEE T rans. Veh. T echnol., vol. 66, no. 10, Oct.
2017, pp. 9534-9549.
[26] Sujata V Mallapur, Siddarama R. Patil, ”Stable Backbone Based
Multipath Routing Protocol for Mobile Ad-Hoc Networks”,
International Conference on Circuits, Power and Computing
T echnologies, 2013, pp 245-248.

978-1-7281-5461-9/20/$31.00 ©2020 IEEE 722

Authorized licensed use limited to: Newcastle University. Downloaded on October 18,2020 at 15:38:30 UTC from IEEE Xplore. Restrictions apply.