Hybrid intrusion detection model for hierarchical wireless sensor network using federated learning

IAES International Journal of Artificial Intelligence (IJ-AI)
Vol. 14, No. 1, February 2025, pp. 492~499
ISSN: 2252-8938, DOI: 10.11591/ijai.v14.i1.pp492-499  492
Journal homepage: http://ijai.iaescore.com
Hybrid intrusion detection model for hierarchical wireless
sensor network using federated learning
Sathishkumar Mani1
, Parasuram Chandrasekaran Kishoreraja2
, Christeena Joseph3
, Reji Manoharan4
,
Prasannavenkatesan Theerthagiri1
1
Department of Computer Science and Engineering, GITAM School of Technology, GITAM University, Bengaluru, India
2
School of Computer Science and Information Systems, Vellore Institute of Technology, Vellore, India
3
Department of Electronics and Communication Engineering, SRM Institute of Science and Technology, Chennai, India
4
Department of Electronics and Communication Engineering, Rohini College of Engineering and Technology, Kanyakumari, India
Article Info ABSTRACT
Article history:
Received Apr 18, 2024
Revised Aug 12, 2024
Accepted Aug 30, 2024
The applications of wireless sensor networks are vast and popular in today’s
technology world. These networks consist of small, independent sensors that
are capable of measuring various physical quantities. Deployment of wireless
sensor networks increased due to immense applications which are susceptible
to different types of attacks in an unprotected and open region. Intrusion
detection systems (IDS) play a vital part in any secured environment for any
network. IDS using federated learning have the potential to achieve better
classification accuracy. Usually, all the data is stored in centralized server in
order to communicate between the systems. On the other hand, federated
learning is a distributed learning technique that does not transfer data but
trains models locally and transfers the parameters to the centralized server.
The proposed research uses a hybrid IDS for wireless sensor networks using
federating learning. The detection takes place in real-time through detailed
analysis of attacks at different levels in a decentralized manner. Hybrid IDS
are designed for node level, cluster level and the base station where federated
learning acts as a client and aggregated server.
Keywords:
Attacks
Federated learning
Global aggregator server
Intrusion detection systems
Wireless sensor network
This is an open access article under the CC BY-SA license.
Corresponding Author:
Parasuram Chandrasekaran Kishoreraja
School of Computer Science and Information Systems, Vellore Institute of Technology
Vellore, India
Email: kishoreraja.pc@vit.ac.in
1. INTRODUCTION
The wireless sensor network is extensively used in many applications like measuring the temperature
of volcanoes. The network is simple, low-cost, energy-efficient, and distributed sensing and processing, which
depicts the network to security attacks [1]. Traditional methods like cryptography methods are no longer used
to defend the network. Instead, an intrusion detection system (IDS) detects all kinds of attacks. Considering all
limitations in wireless sensor networks, IDS are specific to detect particular types of attacks [2]. Most attacks
happened in sensor networks due to misbehavior of route updates. This research work uses different levels of
IDS using federated learning. Most of the IDS use a distributed detection process in order to lessen the
computational load, but another problem arises: communication overhead [3], [4]. The authors presented
anomaly detection and communicated to a global model system. It uses estimators for its anomaly IDS [4].
There are various classifiers, co-variance parameters, and statistical tools used to [5]–[9] detect distributed
anomalies. Hybrid algorithms are proposed using Quantum particle swarm optimization (PSO) and radial basis
function neural network (RBFNN) [10]–[12]. Numerous neural network-based IDS are proposed, which give
better approximation ability, good classification and fast convergence [13].

Int J Artif Intell ISSN: 2252-8938 
Hybrid intrusion detection model for hierarchical wireless sensor network … (Sathishkumar Mani)
493
In recent years, machine learning (ML) and deep learning (DL) algorithms have been used in many
domains, such as healthcare and image processing. IDS, using ML techniques, learns all kinds of traffic [14],
[15]. The detection process is totally based on data collected and stored centrally in the server. It is found that
the accuracy decreases due to large data sets with high packet loss rates [16], [17]. The problems can be
addressed by federated learning. This algorithm learns data generated by the devices in a collaborative fashion
without any centralized server. It works with decentralized data from devices which are communicated in either
direction. The traditional centralized learning methods expect to follow local learning and attainment of privacy
preservation and cost reduction [18]–[20]. This research proposed hybrid intrusion detection for wireless sensor
networks using federated learning. The authors proposed a DL-based IDS with four different strategies [21].
Kwon et al. [20] reviewed only seven DL-based solutions for IDS.
There are different types of IDS available, which evaluate various information available on single or
multiple hosts as well as analyzing from captured packets during transmission between the nodes. Signature-
based intrusion detection uses patterns in the detection model, whereas the anomaly detection model looks for
abnormality in network traffic. Anomaly detection techniques use statistical models, neural networks, data
mining, and computational intelligence in the learning module. Today, DL models and artificial intelligence
techniques gather a lot of interest in designing the intrusion detection model. One of the problems is feature
selection, which affects the entire performance of the systems. There is a tradeoff between security and
performance metrics while choosing an IDS technique.
There are different types of IDS for wireless sensor networks with respect to architectural design. One
is centralized, and the other is distributed [14]. Today, most of the IDS are distributed where the detection is
done in a local node. The problem is to spend a significant energy for coordination among all nodes. Further,
the nodes are unable to detect certain attacks since it has knowledge about its neighborhood only. Single point
failure occurs in centralized IDS due to communication problems between the nodes that create large
communication overhead. The third type of IDS is the hybrid model, which is a combination of distributed and
centralized IDS.
Learning algorithms like ML and DL are used in many domains, such as healthcare and image
processing. IDS, using ML techniques, learns all kinds of traffic. The detection process is totally based on data
collected and stored centrally in the server. It is found that the accuracy decreases due to a large data set with
a high packet loss rate [20]. The problems can be addressed by federated learning. This algorithm learns data
generated by the devices in a collaborative fashion without any centralized server [22]. It works with
decentralized data from devices which are communicated in either direction. There are two stages: local
learning and model transmission, which permit the accomplishment of privacy preservation and cost reduction.
In the traditional method of intrusion detection, all information is maintained in a centralized server and also
transferred this information between server and host, which are vulnerable to man-in-the-middle attacks [23].
Federated learning methods work in a decentralized manner with this information [24]. So, it is efficient and
enforces a privacy policy for sensitive data. There are numerous approaches to intrusion detection using
federated learning. Sunny et al. [25] proposes using mimic learning in combination with federated learning to
protect against reverse engineering attacks. Most ML and DL models suffered from false negative
alarms [26], [27].
This research proposed hybrid intrusion detection for wireless sensor networks using a federated
learning algorithm. A typical artificial neural network has different phases. It uses supervised and unsupervised
training algorithms. The pattern recognition problems can be solved by incorporating the supervised
algorithms. The classification problems can be solved by incorporating unsupervised algorithms where the
network learns without the knowledge of the desired output [28], [29]. The significance of the neural network
is that it repeatedly learns the coefficients. The coefficients are adjusted to normal data and attack data during
the training phase. The neural network approach improves the detection rate, and the false alarm is
reduced [30], [31].
2. METHOD
The proposed model is layered and clustered with a hybrid IDS model. Each IDS is placed on the
client side with sensor nodes. All are operated in a distributed manner. A hybrid Hierarchical network consists
of a sensor node and a client which holds local IDS systems. It identifies the data anomaly with respect to the
sensor node and client. Anomaly is detected based on the mean variance and data distribution is correlated with
sensor node and client locally. After the selection of cluster heads among nodes, cluster-based IDS (CBIDS)
is activated in order to detect different types of attacks such as selective forwarding, flooding, selfish
misbehaviour, node replication attacks and sinkhole attacks. All behaviour analysis is done using federated
learning architecture. The federated learning-based IDS for wireless sensor network architecture is given in
Figure 1.

 ISSN: 2252-8938
Int J Artif Intell, Vol. 14, No. 1, February 2025: 492-499
494
Figure 1. Federated learning-based IDS in wireless sensor network
The proposed architecture is divided into two parts—the client side and the server side. The
client-side has local model data aggregation and sensor devices. The server side has a global aggregator server
and a local model aggregator, which leads to the global model. In this architect, each local client model trains
the data acquired from sensor devices with the local models shared by the server. Further, the IDS at the client
end detects any unwanted attacks at the node level and cluster level. An analyzer is used to monitor and track
their network traffic data as well as node data for subsequent analysis. Clients are trained locally and globally
for data aggregation. The detection module aggregates all trained data from clients and analyse the data to
check the abnormal behavior of the wireless sensor network. The use of a global aggregation server is to
transform local learning into global learning. A client is able to detect intrusions by comparing behaviors
obtained from global learning and improves the detection.
2.1. Implementation
The proposed work first client trains a local dataset and then shares the data with a global aggregator
rather than on a central server. The global aggregation server interacts with all clients and looks for local IDS
models. It creates an updated global model with all client's IDS models with optimal parameters. The equation
uses the starting weights (w) and number of federated learning rounds (R); the convergence level can be
achieved by changing weights and number of federated learning rounds again and again. At round t, each local
client’s weight is communicated and updated to the aggregation server (1) is used from the FedAvg algorithm
[28] to update the model weights.
𝑊𝑡+1= ∑𝑘=1
𝐾 𝑛𝑘 / 𝑛 𝑤𝑘
t+1 (1)
Where 𝑛 is the total size of all client datasets, and 𝑛𝑘represents the size of each client dataset. 𝑤𝑘
t+1 is the
updated global model after the iteration.
2.2. Algorithm for local intrusion detection system on the client side
The algorithm for the local IDS on the client side is summarized in this section. The following
algorithm is implemented on each client side, and the local analyzer evaluates sense data for abnormalities.
Step 1: Check the sensor data. create a table and store it
Step 2: Take the table. Check the size and compare it with a threshold. Compute variance
Step 3: Compute abnormalities in the table. Check the condition of data anomaly with a threshold value
Step 4: Otherwise, drop it. Forward to a global leader
The global aggregation server initiates a neural network model (NNM) from a global intrusion
detection model. Each uses the global model. Each client creates local weights with their private data, and each
client calculates a fresh set of local weights and works parallelly with the global model. The clients use sensor
data collected locally and analyse with local analyser. The local client model works with the global aggregation
model in order to improve the IDS and communicate with the global aggregation server. The global aggregation
server adapts changes received from local clients and adds the weights from the various local node models in
order to produce a new, improved model (1). The parameters are evaluated on the basis of dataset size at every
node. Once again, updated model parameters are communicated with clients for the changes that occurred in

495
the centralized server. Every client routine the novel classical parameters and makes variations to them based
on the novel data. The process is repeated for the improvement of learning.
Data packet information is given in the input of a neural network, which has one input layer, one
hidden layer and an output layer. This neural network-based IDS is implemented in cluster head for detecting
the attacks. So, four neurons have been given to the input of the neural network. Normal and abnormal
condition is created in the training phase. Centre of activation function and spread factor is initiated and the
spread factor. Performance is totally based on a number of parameters in a neural network. Hidden layer
parameters and radius of the RDF function play crucial functions in the performance of the IDS system.
2.3. Algorithm for global intrusion detection system
The algorithm for the global IDS is summarized in this section. The input parameters to the federated
learning structure are given. Different types of attacks can be detected.
Step 1: Initialize self-organisation map parameters
Step 2: Initialize the weights for the neural network
Step 3: Compute the output of every node and error Function4.
Step 4: Check the condition for error
Step 5: Otherwise, update the self-organization parameters and weights of the neural network. Calculate the
output of every node.
The detection of attack and techniques is tabulated in Table 1.
Table 1. Detection and techniques
IDS Technique
CBIDS Federated learning methods
SBIDS Federated learning methods
NBIDS Rule-based
3. RESULTS AND DISCUSSION
Tensorflow and keras are used for ML and DL. Simulation is carried out in network simulator
version 2 (NS2) in order to extract wireless sensor network parameters. Simulation parameters are listed in
Table 2. Datasets are generated from NS2 and sensors to train and test IDS. The testbed was created for two
types of attacks which are distributed denial-of-service (DDoS) attacks and man-in-the-middle (MIM) attacks.
Initially, the data is preprocessed. The feature selection approaches were applied to reduce training and
classification time. Table 3 displays the selected data for ML models. The tests were performed using federated
learning. The efficacy of federated learning was evaluated with 2 or 3 clients.
Table 2. Simulation parameters
Parameters Values
Area 600×600
Nodes (Number) 100
Simulation time in seconds 100
Protocol for routing HIDS
Energy (Joules) 100
Interval 4 to 6
Number of attackers - 4
Packet Size (bytes) 50 to 100
Table 3. Data set for ML model
Type Total Train Test
Normal 11,222 8,856 2,466
DDoS_UDP Attack 5,508 4,478 3,299
DDoS_ICMP Attack 8,195 6,989 3,956
DDoS_HTTP Attack 9,789 7,221 3,777
DDoS_TCP Attack 8,358 6,136 3,546
MIM Attack 1,725 1,238 674
3.1. Performation evaluation
Network performance parameters were analyzed with different packet size e and interval.
Performance is measured with two parameters. They are the detection ratio and false positive rate. The graphs
are analyzed in Figure 2. Figure 2(a) depicts the packet size vs jitter, Figure 2(b) depicts the packet size vs

 ISSN: 2252-8938
496
packets dropped, Figure 2(c) depicts the interval vs packets dropped, Figure 2(d) depicts the interval vs
throughput, Figure 2(e) depicts the attacker vs throughput, and Figure 2(f) depicts the jitter vs attacker. The
results show an improvement in the performance of the anomaly detection performance system with a reduction
of false positive rate and high detection ratio. It shows the performance graph for hybrid IDS architecture. IDS
performance can be evaluated by the following parameters. Correct positive (CP): the number of attack samples
out is divided by accurately detected attacks in the total samples. Untrue positive (UP): the number of normal
samples is divided by incorrectly identified as attacks in the normal samples. Correct negative (CN): the
number of benign samples is divided accurately and classified as normal. Untrue negative (UN): the number
of attack samples is divided by wrongly recognized as normal.
(a) (b)
(c) (d)
(e) (f)
Figure 2. Performance graph for data anomaly in NBIDS: (a) packet size vs jitte, (b) packet size vs packets
dropped, (c) interval vs packets dropped, (d) interval vs throughput, (e) attacker vs throughput, and
(f) jitter vs attacker

497
The performance metrics are evaluated by two parameters. Detection ratio: the ratio between the
number of correctly identified attacks and expected attacks. False alarm rate: it is the ratio between the
identification of normal samples as attack with normal samples. Figures 3(a) and 3(b) shows the performance
graph for hybrid IDS architecture. Table 4 shows the results of ML approaches for a centralized model in terms
of detection ratio. This table gives information about how IDS differentiates attacks and benign classes in the
dataset. The detection ratio for RNN and CNN approaches is reached at peak values of 93% and 95%,
respectively. Table 5 shows the comparison of the detection ratio in global models.
(a) (b)
Figure 3. Performance graph for hybrid IDS architecture (a) attacker vs false positive and (b) attacker vs
detection ratio
Table 4. Intrusion detection at global aggregation server
Class Detection ratio
CNN RNN
Normal 0.93 0.95
DDoS_UDP Attack 0.88 0.89
DDoS_ICMP Attack 0.80 0.81
DDoS_HTTP Attack 0.60 0.55
DDoS_TCP Attack 0.93 0.94
MIM Attack 0.93 0.95
Table 5 Comparison of detection ratio
Classifier Clients Federated learning model detection ratio
CNN 2 64.23
3 61.88
RNN 2 60.39
3 61.47
4. CONCLUSION
The IDS model is proposed in this research work using federated learning for wireless sensor
networks. Three IDS models have been designed at three levels. One is at the sensor side, the second is at the
client base, and the last is at the global aggregation server side. Three IDS aims to detect different types of
attacks using a federated learning model. It is observed that the hybrid IDS model using federated learning
gives a high detection ratio above 92 %—and a low false positive rate. Further, IDS can achieve a very low
false positive rate by changing the parameters in the federated learning model.
REFERENCES
[1] Y. Maleh and A. Ezzati, “A review of security attacks and intrusion detection schemes in wireless sensor network,” International
Journal of Wireless & Mobile Networks, vol. 5, no. 6, pp. 79–90, 2013, doi: 10.5121/ijwmn.2013.5606.
[2] H. Sedjelmaci and S. M. Senouci, “A lightweight hybrid security framework for wireless sensor networks,” 2014 IEEE International
Conference on Communications (ICC), Sydney, Australia, 2014, pp. 3636-3641, doi: 10.1109/ICC.2014.6883886.
[3] Z. Yang, N. Meratnia, and P. Havinga, “An online outlier detection technique for wireless sensor networks using unsupervised
quarter-sphere support vector machine,” in 2008 International Conference on Intelligent Sensors, Sensor Networks and Information

 ISSN: 2252-8938
498
Processing, 2008, pp. 151–156, doi: 10.1109/ISSNIP.2008.4761978.
[4] S. Rajasegarar, C. Leckie, M. Palaniswami, and J. C. Bezdek, “Quarter sphere based distributed anomaly detection in wireless
sensor networks,” 2007 IEEE International Conference on Communications, Glasgow, UK, 2007, pp. 3864-3869, doi:
10.1109/ICC.2007.637..
[5] N. Shahid, I. H. Naqvi, and S. B. Qaisar, “Quarter-sphere SVM: Attribute and spatio-temporal correlations based outlier & event
detection in wireless sensor networks,” IEEE Wireless Communications and Networking Conference, WCNC, pp. 2048–2053, 2012,
doi: 10.1109/WCNC.2012.6214127.
[6] Y. Zhang, N. A. S. Hamm, N. Meratnia, A. Stein, M. V. D. Voort, and P. J. M. Havinga, “Statistics-based outlier detection for
wireless sensor networks,” International Journal of Geographical Information Science, vol. 26, no. 8, pp. 1373–1392, 2012, doi:
10.1080/13658816.2012.654493.
[7] M. A. Rassam, A. Zainal, and M. A. Maarof, “One-class principal component classifier for anomaly detection in wireless sensor
network,” in 2012 Fourth International Conference on Computational Aspects of Social Networks (CASoN), 2012, pp. 271–276,
doi: 10.1109/CASoN.2012.6412414.
[8] J. Weng, Y. Zhang, and W. S. Hwang, “Candid covariance-free incremental principal component analysis,” IEEE Transactions on
Pattern Analysis and Machine Intelligence, vol. 25, no. 8, pp. 1034–1040, 2003, doi: 10.1109/TPAMI.2003.1217609.
[9] M. A. Rassam, A. Zainal, and M. A. Maarof, “An efficient distributed anomaly detection model for wireless sensor networks,”
AASRI Procedia, vol. 5, pp. 9–14, 2013, doi: 10.1016/j.aasri.2013.10.052.
[10] Y. Liu, “QPSO-optimized RBF neural network for network anomaly detection,” Journal of Information and Computational Science,
vol. 8, no. 9, pp. 1479–1485, 2011.
[11] Y. Peng, Y. Wang, Y. Niu, and Q. Hu, “Application study on intrusion detection system using IRBF,” Journal of Software, vol. 9,
no. 1, pp. 177–183, 2014, doi: 10.4304/jsw.9.1.177-183.
[12] P. Yichun, N. Yi, and H. Qiwei, “Research on intrusion detection system based on IRBF,” in 2012 Eighth International Conference
on Computational Intelligence and Security, 2012, pp. 544–548, doi: 10.1109/CIS.2012.128.
[13] S. Devaraju and S. Ramakrishnan, “Performance analysis of intrusion detection system using various neural network classifiers,”
in International Conference on Recent Trends in Information Technology, ICRTIT 2011, 2011, pp. 1033–1038, doi:
10.1109/ICRTIT.2011.5972289.
[14] M. Riecker, S. Biedermann, R. El Bansarkhani, and M. Hollick, “Lightweight energy consumption-based intrusion detection system
for wireless sensor networks,” International Journal of Information Security, vol. 14, no. 2, pp. 155–167, 2015, doi:
10.1007/s10207-014-0241-1.
[15] U. Halici, “Radial basis function network,” in Artificial Neural Networks, Ankara, Turkey: METU Electrical and Electronics
Engineering, 2005, pp. 139–147.
[16] E. Hodo, X. Bellekens, A. Hamilton, C. Tachtatzis, and R. Atkinson, “Shallow and deep networks intrusion detection system: a
taxonomy and survey,” arXiv-Computer Science, pp. 1–43, 2017.
[17] F. Tang, B. Mao, Z. M. Fadlullah, and N. Kato, “On a novel deep-learning-based intelligent partially overlapping channel
assignment in SDN-IoT,” IEEE Communications Magazine, vol. 56, no. 9, pp. 80–86, Sep. 2018, doi:
10.1109/MCOM.2018.1701227.
[18] C. Zhang, Y. Xie, H. Bai, B. Yu, W. Li, and Y. Gao, “A survey on federated learning,” Knowledge-Based Systems, vol. 216,
pp. 1–11, 2021, doi: 10.1016/j.knosys.2021.106775.
[19] V. Mothukuri, P. Khare, R. M. Parizi, S. Pouriyeh, A. Dehghantanha, and G. Srivastava, “Federated-learning-based anomaly
detection for IoT security attacks,” IEEE Internet of Things Journal, vol. 9, no. 4, pp. 2545–2554, 2022, doi:
10.1109/JIOT.2021.3077803.
[20] D. Kwon, H. Kim, J. Kim, S. C. Suh, I. Kim, and K. J. Kim, “A survey of deep learning-based network anomaly detection,” Cluster
Computing, vol. 22, pp. 949–961, 2019, doi: 10.1007/s10586-017-1117-8.
[21] A. Aldweesh, A. Derhab, and A. Z. Emam, “Deep learning approaches for anomaly-based intrusion detection systems: a survey,
taxonomy, and open issues,” Knowledge-Based Systems, vol. 189, 2020, doi: 10.1016/j.knosys.2019.105124.
[22] B. Cetin, A. Lazar, J. Kim, A. Sim, and K. Wu, “Federated wireless network intrusion detection,” in 2019 IEEE International
Conference on Big Data (Big Data), 2019, pp. 6004–6006, doi: 10.1109/BigData47090.2019.9005507.
[23] N. Bouacida and P. Mohapatra, “Vulnerabilities in federated learning,” IEEE Access, vol. 9, pp. 63229–63249, 2021, doi:
10.1109/ACCESS.2021.3075203.
[24] A. Thakkar and R. Lohiya, “A review on machine learning and deep learning perspectives of IDS for IoT: recent updates, security
issues, and challenges,” Archives of Computational Methods in Engineering, vol. 28, no. 4, pp. 3211–3243, 2021, doi:
10.1007/s11831-020-09496-0.
[25] A. I. Sunny, A. Zhao, L. Li, and S. K. Sakiliba, “Low-cost IoT-based sensor system: A case study on harsh environmental
monitoring,” Sensors, vol. 21, no. 1, pp. 1–12, 2021, doi: 10.3390/s21010214.
[26] N. A. A. Al-Marri, B. S. Ciftler, and M. M. Abdallah, “Federated mimic learning for privacy preserving intrusion detection,” in
2020 IEEE International Black Sea Conference on Communications and Networking (BlackSeaCom), 2020, pp. 1–6, doi:
10.1109/BlackSeaCom48709.2020.9234959.
[27] M. A. Ferrag, O. Friha, D. Hamouda, L. Maglaras, and H. Janicke, “Edge-IIoTset: a new comprehensive realistic cyber security
dataset of IoT and IIoT applications for centralized and federated learning,” IEEE Access, vol. 10, pp. 40281–40306, 2022, doi:
10.1109/ACCESS.2022.3165809.
[28] H. B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y Arcas, “Communication-efficient learning of deep networks from
decentralized data,” Proceedings of the 20th International Conference on Artificial Intelligence and Statistics, AISTATS 2017, vol.
54, pp. 1-10, 2017.
[29] T. D. Nguyen, S. Marchal, M. Miettinen, H. Fereidooni, N. Asokan, and A.-R. Sadeghi, “DÏoT: a federated self-learning anomaly
detection system for IoT,” in 2019 IEEE 39th International Conference on Distributed Computing Systems (ICDCS), 2019, pp.
756–767, doi: 10.1109/ICDCS.2019.00080.
[30] R. Samrin and D. Vasumathi, “Review on anomaly based network intrusion detection system,” in 2017 International Conference
on Electrical, Electronics, Communication, Computer, and Optimization Techniques (ICEECCOT), 2017, pp. 141–147, doi:
10.1109/ICEECCOT.2017.8284655.
[31] P. V. Theerthagiri and M. Thangavelu, “Elephant intrusion warning system using IoT and 6LoWPAN,” International Journal of
Sensors, Wireless Communications and Control, vol. 10, no. 4, pp. 605–616, 2019, doi: 10.2174/2210327909666191129092006.

499
BIOGRAPHIES OF AUTHORS
Sathishkumar Mani has obtained his B.E. degree in Computer Science and
Engineering from Bharathiar University, Coimbatore, India and M.Tech. degree in Information
Technology from Punjabi University, Patiala, India. He earned his Ph.D. in Computer Science
and Engineering from Saveetha University, Chennai, India. He has over 25 years of experience
in multiple domains like teaching, research and software development. His research area is
network security, machine learning, and IoT. He can be contacted at email:
sathishkumarmani17@gmail.com.
Parasuram Chandrasekaran Kishoreraja is a Professor at the School of
Information Technology in Vellore Institute of Technology (VIT), Vellore, India. He has a total
experience of 23 years in academic and research. He has published and presented various papers
in the international journal and conferences. His research interests include security systems,
internet of things, and medical artificial intelligence. He can be contacted at email:
kishoreraja.pc@vit.ac.in
Christeena Joseph is an Associate Professor working in the Department of ECE
at SRM Institute of Science and Technology, Ramapuram, Chennai, India. She has teaching
experience of 17 years and has published 50 research papers in national and international
journals. Her research interests include wireless communication and networks. She can be
contacted at email: christeena003@gmail.com.
Reji Manoharan is an Associate Professor in the Department of Electronics and
Communication Engineering at Rohini College of Engineering and Technology Kanyakumari
India. He has a total experience of 15 years in academic and research. He has published and
presented various papers in the international journal and conferences. His research interests
include network security, internet of things, antenna design, and intrusion detection. He can be
contacted at email: rejieceped@gmail.com.
Prasannavenkatesan Theerthagiri is working as the Assistant Professor in the
Department of Computer Science and Engineering, GITAM Deemed to be University,
Bengaluru, India. He was awarded Ph.D. (Full-Time) degree in the year 2021 on the work of
wireless communication with machine learning from Anna University, Chennai, India. He was
awarded the mobility grant award by the Republic of Slovenia in the year 2017-2018. He has
published his research works in 12 SCI indexed journals, 16 SCOPUS indexed journals. His
research interests are data science, AI, IoT, and MANET. He can be contacted at email:
prasannait91@gmail.com.

Hybrid intrusion detection model for hierarchical wireless sensor network using federated learning

More Related Content

Similar to Hybrid intrusion detection model for hierarchical wireless sensor network using federated learning (20)

More from IAESIJAI (20)

Recently uploaded (20)

Hybrid intrusion detection model for hierarchical wireless sensor network using federated learning