A Securing The Manet by Detecting The Flooding Attacks Using Hybrid CNN-Bi-Lstm-Rf Model

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 14

9th International Conference, Mike

2023, Kristiansand, Norway


Volume 7, Issue Xii, Issn No: 1236- 6124.

A Securing The Manet By Detecting The Flooding Attacks


Using Hybrid CNN-Bi-Lstm-Rf Model
*B.Deena Divya Nayomi
Assistant Professor,Department of CSE,
G.Pullaih college of engineering &Technology College,
Kurnool, deenadivyanayomecse@gpcet.ac.in.
J.Siva ani
Assistant Professor, Department of IT,
SRM Institute of Science and Technology, Chennai,
Sivarani @srmist.edu.in
S. Vatchala
Research Scholar, Anna University, Chennai
vatchalacse@gmail.com
B.Sowjanya
Research Scholar, Anna University, Chennai,
Sowjanya.b@gmail.com
Abstract
Issue Details
Issue Title: Issue 1 Over the years, ensuring security over Mobile Ad-hoc (MANET) Routing has been a
Received: 08 February, 2023
Accepted: 19 March, 2023 challenging task for both the researchers and academicians due to its natural and unique
Published: 22 June, 2023 characteristics of mobility, self organizing and free to join/leave features. Generally, an
Pages:- 620-633 intruder can become part of MANET and can broadcast unnecessary or useless packets over
the network to disrupt network activities by increasing network overhead, consuming
network bandwidth and deploy the network with battery energy consumption. In the past,
many research works were focused and dedicated to address this flooding attack issue. In
this work, we propose machine learning based algorithm to detect flooding attacks that
considers route discovery history information of each nodes and use it to identify the
malicious node if the characteristics deviates from this class information. This work also
introduces Enhanced-AODV for preventing flooding attacks (EFAP-AODV) to ensure the
network security by effectively detecting and isolating malicious nodes from participating
network functionalities. The obtained results were compared with AODV, B-AODV and
our proposed approach outplays these two approaches in terms of packet delivery ratio, end
to end delay, routing load and malicious node detection ratio.

Keywords: Mobile Ad-hoc Network, Secure Routing, Machine Learning, k-NN algorithm,
Flooding attack prevention.
Introduction:

A Mobile Ad-Hoc Network consists of collection of wireless devices (referred as nodes) that dynamically
forms a network with no centralized infrastructure to offer communication among the nodes. In general, these nodes
operate with limited battery power and dynamic topology where every node in the network can act as a host and
router. These networks are self-configured due to its mobility nature and can be rapidly deployed and reconfigured.
MANETs are commonly used in emergency situations, natural disaster and rescue scenarios, healthcare, wildlife,
Internet of Things (IoT) and military applications. Basically, mobile devices are succumbed to its design constraints
Size, Width and Portability [1, 2].

Over the years, MANET has consistently attracted researchers due to its unique characteristics, mainly due
to decentralized architecture, dynamic topology and security loopholes. Routing in MANET has always been a
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.
challenging task as it has no physical medium to communicate, moreover, flooding is an important issue as far as
network traffic, energy dissipation and security features are concerned. Identifying malicious node in MANET is
very difficult task unless security algorithms are embedded with routing strategies [3]. To ensure security in
MANET routing, MANET routing strategies should adapt security schemes with more flexibility on the basis of its
ever changing network topology environment. Further, all necessary security attack measures should be addressed
and rectified to improve the overall performance of MANETs in all aspects [4].

Security in MANET:

In MANET, security related problems have been dealt with the usage of common and modified routing
schemes such as reactive routing schemes were effective “for threat identification and faster response time” whereas
table-driven routing schemes were effective “on resisting adversary from invoking intruders” [5]. Over the decades,
the researchers have explored security approaches for MANET in two ways: securing routing schemes [6] and data
security scheme [7]. In [8] the author has presented the prominent reasons behind the security loopholes in MANET
as follows: i) absence of competent and strong resistance from adversary nodes ii) presence of fatal threats within
the network due to the free to join/leave feature iii) absence of centralized architecture iv) freedom of
mobility/fluctuating scalability v) Impact of limited battery power on active/inactive nodes within the network.

Ensuring security in MANET is a complex task since all layers of its communication stack are prone to
unknown problems such as hidden node problem and exposed node problem. In few scenarios, unexpected threats
and attacks can take place because links can be created and broken between the nodes at anytime.

Table 1 Security challenges in MANET.

Challenge Description
Wireless Link Open and physically accessible to everyone, prone to bit
error/interference
Lack of secure boundaries Adversaries can easily join and become part of the
network/can leave at anytime
Infrastructure- less No specific infrastructure for addressing, key distribution,
certification etc.,
Nodes Limitation Since the nodes have limited capabilities, their
availability can easily be compromised
Link Limitation Cooperation based security algorithms must consider the
bandwidth limitation associated with the linksRepudiation
Multi-hop Routing Application
Since the nodes Layeron each other for routing
are dependent Viruses
adversaries can generate fabricated routes to create
routing loops, false routes etc.
In MANET, the attacks can further be classified based on the “status of the attacker (internal or outsider),
behavior of the attack (active and passive attack) and purpose of the attack (Eavesdropping, Denial of Service, black

hole and etc.)” Hence, the integrity, availability and confidentiality of the network and data must be protected at all
layers. To tackle or overcome this issue, novel routing schemes need to be adopted at network layer to avoid routing
loops, false routes by effectively identifying the frequent topological changes within the network [9-11].

Physical Layer Physical Layer Signal Jamming

MAC Layer
Link Layer Link Layer disruption and
adversarial attack
Eavesdropping
Traffic
Wormhole, Black
Analysis and
Hole, Byzantine,
Monitoring
Network Layer Resource Table
Network Layer
Overflow, Cache
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.
Passive Attacks Active Attacks

Figure 1.Taxonomy of attacks in MANET [13]

In MANET, routes can be discovered by flooding route request packets over the network and the malicious
node in the network tries to exhaust the network resources and disrupts its operations by broadcasting false packets.
Although, MANETs prone to various attacks, flooding is considered as one of the most lethal attack as few nodes
might appear to be normal during route discovery phase, but become malicious (dropping some packets) during data
forwarding phase of the communication and such an attack is called as smart gray hole attack. Flooding attacks can
be either continuous or selective. Figure 1 depicts the taxonomy of various attacks at different layers of MANET.

Machine Learning:

With ever growing technology and its increased usage among the users across the world in every
imaginable field, privacy and security concerns are exponentially growing with the new range of threats and attacks.
Six out of
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.

seven billion people use their mobile gadgets for their everyday activities ranging from shopping, banking, social
media activities, healthcare and professional activities. This has increased the chances for information leakage and
theft [14-16]. Security problems are at large in MANET applications and its operations even with the availability of
various dedicated approaches. In the recent years, Machine Learning Algorithms (MLA) has played a vital role in
various research fields such as computer vision, wireless networks, wireless sensor networks, data science and etc.
MLA has widely been successful in addressing current security issues and fulfilling the real world requirements with
real time decision making, big data processing, reduced cycle time for learning and error free processing [17].

Various machine learning algorithms has been used to ensure security over MANETs such as deep
learning, support vector machines, k-NN clustering, decision trees, naïve bayes and among them random forest has
yielded significant results when compared to other methods in terms of accuracy, complexity and memory usage
[18].

Literature study:

In [19] Gada has proposed a static threshold method where the nodes have fixed limits on route request
acceptance, route request blacklist and route request rate. However, this approach is not suitable for dynamic
environments as the main characteristic of MANET is its ever changing dynamic topology. In addition, this
approach has suffered with excessive end to end delay and normal nodes might be identified as malicious node due
to static threshold values. Thus, this method has resulted in high false positive rates. In [20,28] author has proposed
Support Vector Machine (SVM) based approach where every nodes behavior is forwarded to SVM to determine
whether the node is malicious or not based on the threshold limit. In [21] authors have proposed k-nearest neighbors
based method with no specific algorithms for training datasets, but considers more generic information to form a m-
dimensional vector to represent nodes and their behaviors.

In [22] Balanced-AODV method was proposed where each node in the network had a Balance Index value
(BIV) based on which a malicious node will be identified. Upon receiving a route request packet, each node
compares the route request rate with BIV, if route request rate is higher than BIV, the request is dropped and the
node will be identified as a malicious node. Otherwise, the packet will be forwarded to the neighboring node. The
main drawback of this method is that node movement can neither be predicted nor fixed as some nodes have smaller
movements whereas other nodes might move in faster rate. Fast moving nodes tend to receive more route request
packets and can be identified as malicious though it is a legitimate node. Furthermore, this approach does not have
any confirmation mechanism to identify a particular node as malicious.

In [23] an improved AODV protocol was proposed by integrating Security Mobile Agents (SMA) where
SMA was trained to obtain the minimum time required (Tmin) for route discovery process from sender to destination.
Upon receiving route request packet, a node checks the time T with T min, if T<Tmin, then the packet dropped and the
sender will be identified as malicious node. The value of Tmin can only be relied if no malicious node exists during
the training.

In [24] authors have addressed the issue and proposed Mitigating Gray Hole Attack Mechanism (MGAM)
that employs several special nodes within a network called as Gray Hole Intrusion Detection System (G-IDS) which
can overhear its neighboring nodes transmission with the sequence number, and if it detects that dropping packets
are greater than the threshold value then it broadcasts ALERT message to entire network about the presence of
malicious node. Then, the node will be blocked from participating in any further communication over the network.
The study has implemented the mechanism in NS 2.35 simulator and the results were significantly better than the
previous methods under smart gray-hole attack. In [25] authors have used both self-organizing map and genetic
algorithm for attack identification in MANET with unsupervised learning by considering the uncertain behavior of
adversaries with the assigned fitness function.

In [26] authors have proposed an Enhanced Average Encounter rate-AODV (EAER-AODV] where 3-
dimensional metric model is formed and the routing path will be selected on the basis of trust value of their
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.
neighboring
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.

nodes and this trust value will be updated frequently based on protocol specification and node’s recommendation.
This model dedicated to offer secure communication with dynamic topology of MANET. The simulation results are
efficient as compared to the existing methods. In [27] author have proposed cluster based trust model for secure
routing using key exchange between neighboring nodes with no certification or central authority to ensure security
of route discovery and data exchange. Clusters were formed and trust agent for each cluster was assigned to check
the trustiness of the nodes within the cluster. If the node is reliable, then the packet is forwarded and this reliability
is decided on the basis of traffic and destination sequence number (DSN). The main drawback of this method was
the communication and trustiness between the clusters was questionable.

Proposed Method:

We consider Ad-hoc On Demand Vector Routing protocol for the proposed work. In the first step, we
define route discovery frequency vector based on node’s history of route discovery. In the second step, we introduce
an algorithm for obtaining training dataset to classify whether a node is normal or malicious one. Finally, we present
flooding attack detection algorithm based AODV protocol. In AODV, route discovery frequencies are within range
for normal nodes, but for abnormal nodes the range will be higher than normal as their intention is to disrupt the
network resources by flooding the network.

Route Discovery Frequency Vector:

The crucial problem in detecting RREQ flooding attacks using k-NN is defining feature vector. As
mentioned earlier in this study, we consider route discovery frequency vector as feature vector with the following
assumptions:

Assumption 1: We define Route Discovery Time (Td) as the time taken between route request broadcast messages
to receive a route response.

Assumption 2: We define Inter-route Discovery Time (Ti) as the time taken between the end of a route discovery to
the start of the next route discovery.

(a) Route Request Flooding (b) HELLO and Data Flooding

Figure 2 Sample Flooding Attacks in MANET

The network scenario aimed to create two set of vector classes, namely, Normal Vector Class (NVC) and
Malicious Vector Class (MVC) and network setup parameters for training dataset is given in figure 4.

Improved k-NN Algorithm:


9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.

An improved k-NN algorithm helps to detect and prevent flooding attack during the route discovery process
of our network. The steps are as follows:

Step 1: When source nodes broadcast route request, all other nodes in the network collect information of source nodes

Step 2: Each node employs route discovery frequency vector (V DFV), upon receiving RREQ message and also uses
machine learning algorithm (obtained from training phase) to determine whether the source node is malicious or
normal.

Begin

Source Node (NS) initiates Route


Discovery Process

NN process the
RREQ using RDH of

Is RDH is

Node NN uses Enhanced-AODV to


classify NS using VDFV

If NN is not a If
neighbor of NS,

NS is malicious node and


terminate algorithm
NN saves
broadcast_id and
If NN is destination or has a route toward the
src_add values into
destination, it unicasts a RREP packet back to its
its cache and adds a neighbor from which it received the RREQ packet
reverse route to (NS); otherwise, it rebroadcasts the RREQ packet.
source node into its
routing table

Figure 3 Work Flow Diagram of EFAP-AODV approach

Step 3: k-NN algorithm classifies route discovery frequency vectors as whether Normal Vector Class or Malicious
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.
Vector Class.

Step 4: Euclidean distance is used to calculate the distance between VN and VM and the formula is as follows:

𝑑(𝑉𝑁, 𝑉𝑀 ) = √∑𝑀 (𝑉𝑁[𝑖] − 𝑉𝑀 [𝑖])2 (1)


𝑖=1

Performance Analysis of AODV, BAODV and EFAP-AODV:

To evaluate the undertaken approaches along with the proposed approaches under route request flooding
attacks, the following parameters were used to configure our network. The maximum vector size is 50 and we
adopted random waypoint model to determine the mobility of nodes at various speeds (10 m/s, 20 m/s and 30 m/s)
and we had setup 2 malicious nodes within the network where each of them flood 20 packets per second and these
are part of 20 pair of source nodes that communicates over the network. Figure 3 represents the network parameters
used to configure our network for this study.

Simulation Tool Used Network Simulator Version 2.35


Number of Nodes 100 normal nodes and 1 malicious
Network Area Size node 2000 m * 2000m
Mobility Model Used Random Way Point
Mobility Speed Normal Nodes:
0m/s, 10m/s, 20m/s, 30m/s and 40m/s
Malicious Node:
Positioned at the Center (100m *

Protocol Used 1000m) User Datagram Protocol


Number of Connections (UDP)
Bit Rate Used 50 UDP Connections
First Data Source Constant Bit Rate
Commences Other Data Traffic At time 0
Sources Commence seconds
Malicious Node Flooding 5 Seconds after the first
commencement
Every Second but varies randomly

Figure 4 Network Design

Parameters Packet Delivery Ratio:

In the absence of malicious nodes, AODV produces better packet delivery ratio with an average of 85.37
%, but its numbers dramatically falls in the presence of malicious nodes, 24.26% and 10.45% in the presence of one
and two malicious nodes respectively. When malicious nodes broadcast fake route request messages, AODV fails to
detect it which damages the network functioning due to high bandwidth consumption and buffer overloads in the
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.
neighboring or intermediate nodes. Similarly, B-AODV does well in normal network scenarios with an average of
83.06% whereas the average drops to 59.12 % and 58.43% with one and two malicious nodes respectively.
Compared to these two approaches, EFAP-AODV achieves packet delivery ratio average of 82.24% when the
attackers employs one or two malicious nodes to disrupt the network functioning. Figure 5 depicts the graphical
representation of AODV, B-AODV and EFAP-AODV based on packet delivery ratio.

90.00
80.00
70.00
60.00
AODV
Packet Delivery

50.00
40.00 BAOD
Ratio (%)

30.00 V
20.00
10.00
0.00
0 1 2 0 1 2 0 1
2

1-10 1-20 1-30


m/s Number of m/s
Of Malicious
Nodes
Figure 5 Packet Delivery Ratio comparisons of AODV, B-AODV and EFAP-AODV
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.

5.000
4.500
4.000
3.500
3.000
AOD
2.500
V
Delay (sec)
End to End

2.000
1.500
1.000
0.500
0.000
0 1 2 0 1 2 0 1 2
1-10 1-20 1-30
m/s
Number mo/fs m/s
Malicious Nodes

Figure 6 End to End Delay comparisons of AODV, B-AODV and EFAP-

AODV End to End Delay:

Although, AODV results in better end to end delay under normal scenarios (0.563s), results in higher end
to end delay with the presence of malicious nodes, 3.031s and 3.674s with one and two malicious nodes
respectively. In the same way, B-AODV results in higher end to end delay than AODV in normal scenario (1.068s),
but returns less value with the presence of one and two malicious nodes 1.054s and 1.142s respectively. This is due
to its ineffectiveness in detecting flooding attacks that result in poor packet delivery ratio and higher end to end
delay. Our proposed method, results an average end to end delay in all scenarios: normal (0.621s), one malicious
node (0.663s) and two malicious node (0.696s) respectively. Figure 6 depicts the graphical analyses of ADOV, B-
AODV and EFAP- AODV in terms of end to end delay.

Routing Load:

In general, routing load will be minimum in the absence of malicious nodes in a network, but can be
ruthless depending on intruder’s intention. In normal network scenario, average routing load with high mobility for
AODV was at about 5.76 packets and the numbers become high with the presence of malicious nodes 178.2
packets and
674.32 packets for one and two malicious nodes respectively.
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.

1000.00

800.00

600.00
AODV
Routing Load

400.00
BAOD
(Packet)

V
200.00

0.00
0 1 2 0 1 2 0 1 2

1-10 1-20 1-30


m/s Number mo/fs m/s

Malicious Nodes

Figure 7 Routing Load comparisons of AODV, B-AODV and EFAP-AODV

As for as routing load is concerned, B-AODV yielded following numbers than AODV in all network scenarios:
normal (3.36 packets), one malicious node (4.56 packets) and two malicious nodes (6.53 packets). EFAP-AODV’s
average routing load under the normal scenario is (5.58 packets), one malicious node (7.24 packets) and two
malicious nodes (8.26 packets) respectively.

As far as routing load is concerned, AODV performs well with high correct detection rate and low mistake
rate, yet EFAP-AODV drops more number of route request packets due to mistake detection. Overall, EFAP-AODV
and AODV offered better performance than B-AODV in terms of Routing load. Figure 7 depicts the graphical
representation of AODV, B-AODV and EFAP-AODV routing load performances.
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.

90.00
80.00
70.00
Detection Ration (%)

60.00
50.00
Malicious Node

AOD
40.00
V
30.00
20.00
10.00
0.00
0 1 2 0 1 2 0 1 2

1-10 1-20 1-30


Nmu/smber of Mmal/iscious
Nodme/ss with varying
vector size from 1 t0 50

Figure 8 Malicious node detection ratio of AODV, B-AODV and EFAP-AODV

Malicious Node Detection Ratio:

Figure 8 represents the malicious node detection rate of the undertaken protocols in the given network
scenario with varying vector size and mobility speed. It is clearly evident from the graphical representation that
AODV and EFAP-AODV are effective in identifying and detecting malicious nodes up to mobility speed of 20s,
then EFAP- AODV overtakes AODV with higher detection rate with high mobility speed of 30s. Thus, our proposed
approach is effective in detecting the malicious node than other two approaches.

Conclusion:

In this research work, we aimed to detect malicious nodes that impose flooding attack over Mobile Ad-hoc
Network (MANET). To tackle this problem, we proposed route discovery frequency vector and improved k-NN
algorithm to detect and isolate the malicious nodes from participating network related activities. Then, we
introduced, Enhanced AODV with Flooding Attack Detection (EFAP-AODV) to prevent malicious nodes with high
mistake detection. The results have shown that our proposed approach yield better performances as compared with
AODV and B-AODV approaches in terms of misbehaving detection ratio, packet delivery ratio, end to end delay
and routing load performance metrics. .

References:
9th International Conference, Mike
2023, Kristiansand, Norway
Volume 7, Issue Xii, Issn No: 1236- 6124.

1. J. Hoebeke, I. Moerman, B. Dhoedt, and P. Demeester, “An overview of mobile ad hoc networks:
Applications and challenges,” Journal of the Communications Network, vol. 3, no. 3, pp. 60–66, 2004.
2. R. Pushparaj, M. Dinakaran. A Location Based Energy Aware Routing for Mobile Adhoc Networks.
International Journal of Communication Antenna and Propagation. August 2014. Vol. (4):114-123. Italy.
3. Dharma P. Agrawal, Qing-An Zeng, “Introduction to Wireless and Mobile Systems”, Cengage Learning,
Technology & Engineering, pp. 640, 2015.
4. Iram Nausheen, Dr.Akhilesh Upadhyay, “A Survey on MANETs: Entrusted Security Challenges”,
International Journal of Future Generation Communications and Networking, Vol. 13, No.3, (2020), pp.48-
58.
5. Reddy, P. Narendra, C. H. Vishnuvardhan, and V. Ramesh. "Routing Attacks in Mobile Ad hoc Networks."
International Journal of Computer Science and Mobile Computing 2, no. 5 (2013).
6. Al-Sakib Khan Pathan, Muhammad Mostafa Monowar, Zubair Md. Fadlullah, “Building Next-Generation
Converged Networks: Theory and Practice”, CRC Press, pp. 608, 2013.
7. Brij B. Gupta, “Computer and Cyber Security: Principles, Algorithm, Applications, and Perspectives”,
CRC Press Business & Economics, pp. 666, 2018.
8. Suresh Chandra Satapathy, Siba K Udgata, Bhabendra Narayan Biswal, “Proceedings of the International
Conference on Frontiers of Intelligent Computing: Theory and Applications (FICTA) 2013”, Springer
Science & Business Media, Computers, pp. 564, 2013.
9. Cayirci, E. and C. Rong, Security in Wireless Ad hoc and Sensor Networks. 2009: Wiley Online Library.
10. Rajan Patel and Pariza Kamboj, “A Survey on Contemporary MANET Security: Approaches for Securing
the MANET”, International Journal of Engineering and Technology, ISSN: 0975-4024, Vol. 9, No.1, 2017.
DOI: 10.21817/ijet/2017/v9i1/170901413.
11. Kumar, S., Machha.Narender, and G.N.Ramesh, Security Provision For Mobile Ad-Hoc Networks Using
Ntp & Fuzzy Logic Techniques. Global Journal of Computer Science and Technology, 2010.
12. Vinoth Kumar, V., Arvind, K.S., Umamaheswaran, S., Suganya, K.S (2019), “Hierarchal Trust Certificate
Distribution using Distributed CA in MANET”, International Journal of Innovative Technology and
Exploring Engineering, 8(10), pp. 2521-2524
13. Shivashankar.T.M and Shivakumar S.B., “Insights on Security Improvements and Implications of Artificial
Intelligence in MANET”, Communications on Applied Electronics (CAE), ISSN:2394-4714, Foundations
of Computer Science FCS, New York, USA, Volume.7-No.27, March 2019, pp.1-9.www.caeaccess.org.
14. Lee, Y.S.; Choi, S.S.; Choi, J.; Song, J.S. A Lightweight Malware Classification Method Based on
Detection Results of Anti-Virus Software. In Proceedings of the 2017 12th Asia Joint Conference on
Information Security (AsiaJCIS), Seoul, Korea, 10–11 August 2017; IEEE: Piscataway, NJ, USA, 2017;
pp. 5–9.
15. ESMA. 2016 Annual Report (IEEE). Available online: https://www.ieee.org/about/annual-report.html
(accessed on 2 May 2021).
16. Li, J.; Sun, L.; Yan, Q.; Li, Z.; Srisa-an,W.; Ye, H. Significant permission identification for machine
learning based android malware detection. IEEE Trans. Ind. Inform. 2017, 14, 3216–3225. [CrossRef].
17. Karthikeyan T., Sekaran, K., Ranjith D., Vinoth Kumar V., & Balajee J M. (2019). Personalized Content
Extraction and Text Classification Using Effective Web Scraping Techniques. International Journal of Web
Portals, 11(2), 41–52. doi:10.4018/ijwp.2019070103
18. Lim, T.-S.; Loh, W.-Y.; Shih, Y.-S. A comparison of prediction accuracy, complexity, and training time of
thirty-three old and new classification algorithms. Mach. Learn. 2000, 40, 203–228.
19. Maithili, K., Vinothkumar, V., & Latha, P. (2018). Analyzing the Security Mechanisms to Prevent
Unauthorized Access in Cloud and Network Security. Journal of Computational and Theoretical
Nanoscience, 15(6), 2059–2063. doi:10.1166/jctn.2018.7407
LINGUISTICA ANTVERPIENSIA, 2021 Issue-3
www.hivt.be
ISSN: 0304-2294

20. V.Vinoth Kumar, Ramamoorthy S (2017), “A Novel method of gateway selection to improve throughput
performance in MANET”, Journal of Advanced Research in Dynamical and Control Systems,9(Special
Issue 16), pp. 420-432.
21. W. Li, P. Yi, Y. Wu, L. Pan, and J. Li, “A New Intrusion Detection System Based on KNN Classification
Algorithm in Wireless Sensor Network,” Journal of Electrical and Computer Engineering, vol. 2014,
Article ID 240217, 8 pages, 2014.
22. Praveen Sundar, P. V., Ranjith, D., Karthikeyan, T., Vinoth Kumar, V., & Jeyakumar, B. (2020). Low
power area efficient adaptive FIR filter for hearing aids using distributed arithmetic architecture.
International Journal of Speech Technology, 23(2), 287–296. doi:10.1007/s10772-020-09686-y
23. Vinoth Kumar V, Karthikeyan T, Praveen Sundar P V, Magesh G, Balajee J.M. (2020). A Quantum
Approach in LiFi Security using Quantum Key Distribution. International Journal of Advanced Science and
Technology, 29(6s), 2345-2354.
24. Gurung, Shashi, and Siddhartha Chauhan. "A novel approach for mitigating gray hole attack in MANET."
Wireless Networks24, no. 2 (2018): 565-579.
25. Kumar V., V., Ramamoorthy S., Kumar V., D., Prabu M., & Balajee J. M. (2021). Design and Evaluation
of Wi-Fi Offloading Mechanism in Heterogeneous Networks. International Journal of e-Collaboration,
17(1), 60–70. doi:10.4018/ijec.2021010104
26. Mukherjee, Saswati, Matangini Chattopadhyay, Samiran Chattopadhyay, and Pragma Kar. "EAER-AODV:
Enhanced Trust Model Based on Average Encounter Rate for Secure Routing in MANET." In Advanced
Computing and Systems for Security, pp. 135-151. Springer, Singapore, 2018.

27. Vinoth Kumar V., Ramamoorthy S. (2018) Secure Adhoc On-Demand Multipath Distance Vector Routing
in MANET. In: Mandal J., Saha G., Kandar D., Maji A. (eds) Proceedings of the International Conference
on Computing and Communication Systems. Lecture Notes in Networks and Systems, Springer, Singapore,
24, pp. 49-63
28. Nagarajan, Senthil Murugan, et al. "Feature selection model for healthcare analysis and classification using
classifier ensemble technique." International Journal of System Assurance Engineering and
Management (2021): 1-12.

View publication stats

You might also like