DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008 Int. J. Exp. Res. Rev., Special Vol.

31: 74-84 (2023)

Supervised learning for Attack Detection in Cloud

Animesh Kumar, Sandip Dutta and Prashant Pranav*

Department of Computer Science and Engineering, Birla Institute of Technology, Mesra, Ranchi, Jharkhand, India

E-mail/Orcid Id:
AKS, animeshkumarcse@gmail.com, https://orcid.org/0009-0007-3679-8025; SD, sandipdutta@bitmesra.ac.in, https://orcid.org/0000-0002-3932-
3048; PP, prashantpranav19@gmail.com, https://orcid.org/0000-0002-3932-3048

Article History: Abstract: In this study, we approach a supervised learning algorithm to detect attacks in
th
Received: 29 Mar., 2023 cloud computing. We categorize “Normal” and “Attack” statuses on the dataset. The model
Accepted: 22nd Jun., 2023 evaluation process uses the kappa statistic, the F1-score, recall, accuracy, and precision.
Published: 30thJul., 2023 The system has a very high detection and efficiency rate, with a detection rate of over 99%.
A total of 9594 cases and 44 distinct columns are included in the dataset. The study's
Keywords: results were displayed using a ROC curve and a confusion matrix. This study focuses on
Cloud Attack, Cloud implementing a supervised learning algorithm for detecting attacks in cloud computing
Computing, Machine environments. The main objective is distinguishing between "Normal" and "Attack"
Learning, Supervised
statuses based on a carefully curated dataset. Several metrics, such as the kappa statistic,
Learning, Security
Issues, Support Vector F1-score, recall, accuracy, and precision, are employed to evaluate the model's
Machine performance. The dataset utilized in this research comprises 9594 cases and encompasses
44 distinct columns, each representing specific features relevant to cloud computing
security. Through a rigorous evaluation process, the algorithm demonstrates exceptional
efficiency, achieving a remarkable detection rate of over 99%. Such high accuracy in
identifying attacks is crucial for ensuring the integrity and security of cloud-based systems.
The significance of this study lies in its successful application of a supervised learning
approach to tackle cloud computing security challenges effectively. The model's high
detection rate and efficiency indicate its potential for real-world deployment in cloud-based
systems, contributing to enhanced threat detection and mitigation. These results hold
promising implications for bolstering the security measures of cloud computing platforms
and safeguarding sensitive data and services from potential attacks.

Introduction dedicated to a single organization and can be hosted

Traditional Cloud computing is the on-demand internally or externally. Hybrid Cloud combines public
provisioning of computer resources over the Internet, and private clouds, allowing data and applications to be
including servers, storage, databases, networking, shared between them (Jain and Rajak, 2023).
software, etc., over the Internet (Butt et al., 2023). It can Figure 1 explains the collaborative efforts of
be easily scaled up or down based on demand, allowing cloudlets, brokers, data centers, and services to provide
for flexibility and cost optimization. The responsibility efficient and accessible cloud services. The process
for infrastructure maintenance, security patches, and begins when a user initiates a request for a specific cloud
updates lies with the cloud provider. Amazon Web service or application. The request is forwarded if a
Service, Microsoft Azure, Google Cloud Platform, IBM cloudlet is nearby, enabling faster processing and reduced
Cloud, and Oracle Cloud are popular cloud service latency. Based on the evaluation, the broker selects the
providers. There are three types of cloud deployment most suitable cloud service provider and communicates
models. The public cloud is available for public use. the user's request and service selection to the chosen data
Private Cloud, where Infrastructure and services are center. The data center receives the request, allocates the

*Corresponding Author: prashantpranav19@gmail.com 74

 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)
necessary computing resources, and executes the detection, predictive analytics, etc., done using ML
requested service or application. This collaborative (Kreuzberger et al., 2023; Kwekha-Rashid et al., 2023).
working of cloudlets, brokers, data centers, and services Supervised learning is a machine learning approach that
optimizes the delivery of cloud services, reduces latency, involves training models on labeled data, where each data
enhances user experience, and provides users with point is associated with a known outcome. Supervised
efficient access to a wide range of computing resources learning aims to learn a mapping function that can
and applications. Some of the security attacks are as accurately predict or classify new, unseen instances based
follows. on their features (Yu et al., 2023; Wu et al., 2023).
Regression and Classification are two types of supervised
learning. Classifiers like simple vector machine (SVM)
(Kurani et al., 2023), Random Forest (Bicego et al.,
2023), Decision Tree (Utukuru et al., 2023), Logistic
regression (Wang et al., 2023), Naïve Bayes (Saleh et al.,
2023), Xtreme Gradient Boosting (XGBoost) (Iban et al.,
2023), K-Nearest Neighbour (K-NN) (Mohy-Eddie et al.,
2023) etc. are an example of supervised learning.
(Aldhyani et al., 2022) suggested EDoS attacks in
cloud computing using ML and DL methods using SVM,
Figure 1. Cloud Computing Architecture
KNN, RF, and Deep learning methods.(Arunkumar et al.,
Denial of Service (DoS) Attacks 2023) Proposed that Gannet Optimization Algorithm-
These attacks can disrupt the normal functioning of based hybrid SVM-ELM mitigates attacks in the cloud.
cloud services, leading to downtime, loss of access, and Matlab software is used for simulation using
potential financial and reputational damages for both the CICIDS2017 datasets and Proposed a DL fusion-based
cloud service provider and its customers (George et al., method to solve DDoS issues in cloud computing. More
2023; Gemmer et al., 2023; Clemens et al., 2023; Ashlam optimized DL methods can be deployed for better cloud
et al., 2023; Anita et al., 2023). Side Channel Attacks: detection. (Patel et al., 2022) reviewed the DL method for
These attacks take advantage of the physical attack detection in the network. (Verma et al., 2023)
characteristics, timing, power consumption, etc., of proposed the RepuTE method for DDoS attack detection
systems to extract sensitive information (Yan et al., 2015; in IoT and fog computing. Classifiers will be designed to
Sahi et al., 2017; Agarwal et al., 2019; Gopinath et al., counter future live IoT network traffic attacks; proposed
2023). Man-In-The-Middle Cryptographic Attacks, the a Text-Mining approach for accident analysis in steel
attacker captures or alters the transmitted data, including plants to reduce human involvement in accident-prone
sensitive user credentials, financial information, or areas. Text mining is done in two phases. Four years of
critical application data (Lu et al., 2021; Ha et al., 2022; data are used for model building and its analysis.
Zhang et al., 2019; Sultan et al., 2022). Strong However, there is scope for future enhancements using
Authentication and Access Controls, Data Encryption, different ML classifiers. The author uses SVM, ANN,
Regular Security Updates, and Patching Network NB, K-NN, and RF Classifiers for predicting and
Security (Ma et al., 2023; Gong et al., 2020) measures analyzing injury severity (Sarkar et al., 2020). In future
like Implementing firewalls, intrusion detection & studies, accidents are also predicted using the time series
prevention systems (Radhakishan et al., 2011; Ren et al., method. Some of the other relevant papers (Sarkar et al.,
2020a,b). Moreover, network segmentation protects 2019; Pramanik et al., 2021; Das et al., 2022; Paramanik
cloud networks from unauthorized access and malicious et al., 2022; Bag et al., 2023; Dey et al., 2023).
activities. This research aims to develop and implement
Machine Learning (ML) is a subset of artificial supervised machine learning algorithms to detect
intelligence (Joshi et al., 2023). that involves the intrusion attempts in a cloud network. The objective is to
development of algorithms and models that enable accurately classify instances as either 'Normal' or 'Attack'
computers to learn from data and make predictions. to enhance the security of cloud computing. This work
Algorithms that can automatically learn and improve aims to design and implement an anomaly detection-
from experience instead of relying on instructions. Image based network intrusion detection system for a cloud
and speech recognition, natural language processing computing network that can detect as many potential
(NLP) (Dash et al., 2023; Khurana et al., 2023), anomaly security issues as feasible.
DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
75
 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)
Table 1. Pseudocode for Logistic Regression
function Logistic_Regression_Train(dataset)
weights = initialize_weights ()
bias = initialize bias ()
for iteration = 1 to num_iterations do:
gradients = compute_gradients (dataset, labels, weights, bias)
weights -= learning_rate * gradients
bias -= learning_rate * compute_bias_gradient(gradients)
return weights, bias
function Logistic_Regression_Predict (dataset, weights, bias):
predictions = []
For instance, In the Dataset
prediction = sigmoid (dot product (weights, instance) + bias)
predictions. Append(prediction)
return predictions
function sigmoid(x):
return 1 / (1 + exp(-x))
Research Questions Recall values were derived. The system with 16 GB
Q1. What precautions must a user consider before RAM, 1 TB ROM, and
going for cloud computing?
Q2. How to secure the data while transferring to the
Cloud?
Q3. How to make sure data stored in the Cloud is
secured?

Contribution
This study focuses on the hidden security attacks of
CSPs that affect the quality of services resulting in much
wastage of cloud resources and client money as the cloud
work on a “Pay per basics model.” Detecting cloud
attacks is very difficult as it involves a large set of traffic
in real-time. Our smart model can help separate both
normal and abnormal packets (attacks) using different
classifiers of ML from the network, which is this paper's
main contribution. In this experiment, we use an actual
dataset from the cloud server. The result of this testing is
equated with different standing systems to prove this
system’s durability and efficiency.
Figure 2. Proposed methodology flowchart
Materials and Method
Table 1 includes initializing the weights and bias, MATLAB version R2023 (a) is used to perform these
iterating over the number of iterations, computing experiments. A private cloud was used as the setting for
gradients, updating the weights and bias using the the creation of the dataset. The private cloud
learning rate, and using the sigmoid function for infrastructure was set up with the help of a KVM type-1
prediction. The `dot product` function calculates the dot hypervisor and an Open Nebula (5.12 version) cloud
product of the weights and an instance of the dataset. management platform. On cloud-based virtual machines,
Figure 2 shows the working of the proposed model. a script was run to generate a synthetic workload
The first dataset is processed. The dataset is simulated on replicating the actual cloud model in real-time. We split
MATLAB 2023(a) using supervised Machine learning our dataset in the ratio of 70: 30 during data pre-
classifiers like LG, SVM, DT, RF, XG-Boost, etc. processing. The dataset is then prepared for training and
Parameters such as accuracy, precision, F1 score, and testing by removing duplicates and outliers. We removed
some extra features from the dataset to reduce the time
DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
76
 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)
Table 2. Feature selection for attack status
LAST_POLL rxbytes_slope rxpackets_slope txpackets_slope timesys_slope Status
1604624102 87.8402 27.2996 89.9974 17.8787 Attack
1604624071 87.9098 28.0725 89.9974 18.4349 Attack
1604624041 88.3794 30.3791 89.9974 34.5923 Attack
1604624012 88.0519 29.5388 89.9974 18.4349 Attack
1604623982 87.9098 28.0725 89.9974 18.4349 Attack
1604623952 87.9098 28.0725 89.9969 18.4349 Attack
1604623922 88.0114 29.5388 89.9973 33.6901 Attack
1604623892 88.0519 29.5388 89.9974 18.4349 Attack
1604623862 88.9491 37.7468 89.9971 17.8787 Attack
1604623831 87.9098 28.0725 89.9959 18.4349 Attack
1604623772 87.9546 28.0725 89.9974 18.4349 Attack
1604623742 87.9098 28.0725 89.9970 33.6901 Attack
1604623712 88.0114 29.5388 89.9968 18.4349 Attack
required to process the data. The irrelevant features and being monitored. The values in each column would
unused variables, such as the Time Stamp, Virtual reflect the respective characteristics at that time.
Machine Identity, Unique Domain Identifier, and Domain Table 3 shows the categories of Normal from the
Name, were removed. A distinct dataset, including dataset. In this research, the accuracy of the four different
characteristics, was developed following pre-processing. ensemble classifiers is analyzed and evaluated using the
The subsequent tests were done on this dataset. In total, area under the curve as the metric of choice performance
the dataset contains 9594 cases and 44 different columns. under comparison between imbalanced data and
The first four columns of the table are used to hold oversampled data.
metadata, which includes the epoch time, the virtual
machine ID, the domain name, and the domain identifier.
Two columns provide specific information about the
available network, RAM, and disk space. In this table’s
last column, record whether the target is currently being
attacked or functioning normally. Datasets are trained
using a classifier like SVM, RF, LR, DT, K-NN, XG-
Boost, and NB. Accuracy, Precision, Recall, F1-score,
and Kappa statistics are all evaluated for each model.

Result and Discussion Figure 3. Distributed Targeted Variables

Table 2 and Table 3 show the selected features, such
as the last poll, rxbytes_slope, and txpackets_slope, with Fig 3 explains data Categorization into normal or
their values representing Attack and Normal status, under attack in pie chart form. Attack equals 2306 while
respectively. Normal equals 7288.
LAST_POLL Figure 4 shows a strong correlation between the status
This column represents the date of the last poll, and and the pace at which packets are sent over the network.
rxbytes_slope represents the slope or rate of change of It illustrates the link between features. The present state
received bytes over time. rxpackets_slope represents the is significantly related to the amount of data transmitted
rate of change of received packets over time. over the network. The status positively correlates with the
txpackets_slope: Column represents the rate of change following metrics: the number of network packets
of transmitted packets over time. received per second, the number of network bytes
timesys_slope: This column represents the rate of change received per second, the number of network bytes
of system time over time. Status represents the device transferred per second, the usage of kernel space, and
user space.

DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
77
 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)
Table 3. Feature selection for Normal status
LAST_POLL rxbytes_slope rxpackets_slope txpackets_slope timesys_slope Status
1604455173 88.20650 30.14140 24.3045 89.9850 Normal
1604455142 87.87080 27.34990 15.9061 89.8986 Normal
1604455113 87.88650 27.29960 32.8285 89.9897 Normal
1604455082 87.87600 27.40760 14.2360 89.8741 Normal
1604455055 87.72410 25.82100 22.7510 89.9864 Normal
1604455024 87.71280 25.71000 18.4349 89.9685 Normal
1604454997 88.10170 29.74490 23.1986 89.9864 Normal
1604454962 87.87600 27.40760 12.5288 89.9580 Normal
1604454935 87.72860 25.86640 16.8584 89.9829 Normal
1604454902 87.79740 26.56510 21.8014 89.9818 Normal
1604454580 87.79740 26.56510 32.2756 89.9887 Normal
1604454542 87.87080 27.34990 17.8533 89.9324 Normal
1604454513 87.79740 26.56510 33.6901 89.9907 Normal
Figure 5 explains the statistical technique of achieved an accuracy of 99.04%, better than many other
classifying objects, data points, or clusters based on their proposed models. Our model compares with (Aldhyani et
similarities or dissimilarities. Cluster characteristics and al., 2022; Fazlullah et al., 2023; Sagarkumar, 2023; GSR
differences between clusters can be analyzed to et al., 2023), which have an accuracy of 86.23%,96.53%,

Figure 4. Correlation Heat Map

understand patterns or groupings in the dataset. They can 92%, and 95%, respectively, as shown graphically in
be interpreted to gain insights into the structure of the Figure 7 below. Table 5 shows the assessment of the
data. It helps discover hidden structures, identify similar precision percentage of our work with other proposed
groups, and facilitate further analysis or decision-making models. We achieved a precision of 95.06%, better than
based on the identified clusters. many other proposed models (Fazlullah et al., 2023; Emil
Figure 6 shows Precision, Recall, F1 Score, Accuracy, et al., 2023), having 95.06% and 86.48%, respectively.
and Kappa Statistics for Logistic regression, SVM, Nave Figure 8 shows the graphical representation of our model
Bayes KNN, Grid Search Decision Trees, Random and other existing methods regarding precision
Forest, and XG Boost. percentage.
Table 4 shows the assessment of the accuracy
percentage of our work with other proposed models. We
DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
78
 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)

Figure 5. Cluster Analysis

Figure 5. Performance Metrics on Oversamples Data

Figure 7. Graphical Analysis of Accuracy Model

DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
79
 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)
Table 4. Performance Evaluation on Accuracy
Parameter Table 5. Model comparison in terms of Precision
Reference Accuracy (%) parameter
Our Model 99.04
Reference Precision (%)
Aldhyani et al., 2022 86.23
Our Model 95.06
Aldhyani et al., 2022 89.84
Khan et al., 2023 91.88
Aldhyani et al., 2022 97.54
Aldhyani et al., 2022
Khan et al., 2023 92.56
97.50
Khan et al., 2023 96.53 Khan et al., 2023 93.83
Khan et al., 2023 94.05 Khan et al., 2023 94.74
Khan et al., 2023 91.41 Khan et al., 2023 92.33
Khan et al., 2023 86.72 Khan et al., 2023 91.99
Khan et al., 2023 94.32 GSR et al., 2023 86.48
Khan et al., 2023 95.46 GSR et al., 2023 83.66
Khan et al., 2023 97.69
Khan et al., 2023 98.56 Conclusion
Khan et al., 2023 We present a way of detecting cloud attacks using a
97.37
supervised learning technique and dataset. Our model
Khan et al., 2023 96.33
gives 99.04 % accuracy, so in many practical scenarios, it
GSR et al., 2023 92.00
can be used as discussed below: As cloud computing has
GSR et al., 2023 89.89
emerged as new technological advancement and most
Patel, 2023 85.00 businesses are deploying cloud services to boost their
Patel, 2023 90.00 business, the cloud is becoming increasingly vulnerable
Patel, 2023 89.00 to cryptographic attacks. These attacks can affect the
Patel, 2023 91.00 smooth working of a business and can even lead to
Patel, 2023 92.00 stilling relevant organizational information. Our model
Patel, 2023 93.00 presents a supervised learning technique to detect cloud
Patel, 2023 95.00 attacks with an accuracy of 99.04% and a precision of
95.06%. Classifiers like Logistic regression, simple
vector machine (SVM), Random Forest, Decision Tree,
Naïve Bayes, Xtreme Gradient Boosting (XGBoost), K-

Figure 6. Comparison of Precision among various existing methods

DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
80
 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)
Nearest Neighbour (K-NN), etc. are used in our 106332.
experimental work. The model can prevent a cloud attack https://doi.org/10.1016/j.engappai.2023.106332
if deployed in the actual scenario. In the future, this Bicego, M. (2023). DisRFC: a dissimilarity-based
model can be used to detect specific cloud attacks like Random Forest Clustering approach. Pattern
Cross Site Scripting (XSS) and SQL Injection attacks. Recognition, 133, 109036.
https://doi.org/10.1016/j.patcog.2022.109036
Acknowledgment Butt, U.A., Amin, R., Mehmood, M. (2023). Cloud
The laboratory facilities and support for this study Security Threats and Solutions: A
were provided by the Birla Institute of Technology, Survey. Wireless Pers Commun, 128, 387–413.
Mesra, Jharkhand, India, which the authors gratefully https://doi.org/10.1007/s11277-022-09960-z
acknowledge. Chauhan, N., Kumar, V., & Dixit, S. (2023). To achieve
sustainability in a supply chain with Digital
Conflict of Interest integration: A TISM approach. International
The authors declare no conflict of interest. Journal of Experimental Research and
References Review, 30, 442-451.
Agrawal, N., & Tapaswi, S. (2019). Defense mechanisms https://doi.org/10.52756/ijerr.2023.v30.041
against DDoS attacks in a cloud computing Clemens, V., Schulz, L. C., Gartner, M., & Hausheer, D.
environment: State-of-the-art and research (2023, May). DDoS Detection in P4 Using
challenges. IEEE Communications Surveys & Hyperloglog and Countmin Sketches. In NOMS
Tutorials, 21(4), 3769-3795. 2023-2023 IEEE/IFIP Network Operations and
https://doi.org/1109/COMST.2019.2934468. Management Symposium, pp. 1-6.
Aldhyani, T. H. H., & Alkahtani, H. (2022). Artificial https://doi.org/10.1109/NOMS56928.2023.10154315.
Intelligence Algorithm-Based Economic Denial Das, S., & Sarkar, S. (2022). News media mining to
of Sustainability Attack Detection Systems: explore speed-crash-traffic association during
Cloud Computing Environments. Sensors, 22(13), COVID-19. Transportation Research Record,
4685. MDPI AG. Retrieved from 03611981221121261.
http://dx.doi.org/10.3390/s22134685 https://doi.org/10.1177/03611981221121261
Anitha, P. T., Dibaba, W., & Boddu, R. (2023, May). Dash, G., Sharma, C., & Sharma, S. (2023). Sustainable
Mitigation of Attacks Using Cybersecurity Deep Marketing and the Role of Social Media: An
Models in Cloud Servers. IEEE, In 2023 Experimental Study Using Natural Language
International Conference on Disruptive Processing (NLP). Sustainability, 15(6), 5443.
Technologies (ICDT). pp. 202-205. MDPI AG. Retrieved from
https://doi.org/10.1109/ICDT57929.2023.101508 http://dx.doi.org/10.3390/su15065443
32. Dey, P., Chowdhury, S., Abadie, A., Yaroson, E. V., &
Arunkumar, M., & Kumar, K. A. (2023). GOSVM: Sarkar, S. (2023). Artificial Intelligence-Driven
Gannet optimization based support vector Supply Chain Resilience in Vietnamese
machine for malicious attack detection in cloud Manufacturing Small-and Medium-Sized
environment. International Journal of Enterprises. International Journal of Production
Information Technology, 15(3), 1653-1660. Research.
https://doi.org/10.1007/s41870-023-01192-z https://doi.org/10.1080/00207543.2023.2179859
Ashlam, A. A., Badii, A., & Stahl, F. (2023). Data- Emil Selvan, G. S. R., Ganeshan, R., Jingle, I., & Ananth,
Mining and Hashing to Prevent Application- J. P. (2023). FACVO-DNFN: Deep learning-
Layer DDoS and SQL Injection Attacks. In 2023 based feature fusion and Distributed Denial of
IEEE International Conference on Advanced Service attack detection in cloud
Systems and Emergent Technologies computing. Knowledge-Based Systems, 261,
(IC_ASET), pp. 01-06. 110132.
https://doi.org/10.1109/IC_ASET58101.2023.10150694. https://doi.org/10.1016/j.knosys.2022.110132
Bag, S., Golder, R., Sarkar, S., & Maity, S. (2023). Gemmer, D. D., Meyer, B. H., de Mello, E. R., Schwarz,
SENE: A novel manifold learning approach for M., Wangham, M. S., & Nogueira, M. (2023,
distracted driving analysis with spatio-temporal May). A Scalable Cyber Security Framework for
and driver praxeological features. Engineering the Experimentation of DDoS Attacks of Things.
Applications of Artificial Intelligence, 123,
DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
81
 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)
In NOMS 2023-2023 IEEE/IFIP Network https://doi.org/10.1109/JAS.2023.123657.
Operations and Management Symposium, pp. 1- Joshi, A., Capezza, S., Alhaji, A., & Chow, M. Y. (2023).
7. Survey on AI and Machine Learning Techniques
https://doi.org/10.1109/NOMS56928.2023.10154400. for Microgrid Energy Management
George, A. S., & Sagayarajan, S. (2023). Securing Cloud Systems. IEEE/CAA Journal of Automatica
Application Infrastructure: Understanding the Sinica, 10(7), 1513-1529.
Penetration Testing Challenges of IaaS, PaaS, https://doi.org/10.1109/JAS.2023.123657.
and SaaS Environments. Partners Universal Khan, F., Jan, M. A., Alturki, R., Alshehri, M. D., Shah,
International Research Journal, 2(1), 24-34. S. T., & ur Rehman, A. (2023). A Secure
https://doi.org/10.5281/zenodo.7723187 Ensemble Learning-Based Fog-Cloud Approach
Gong, S., Ochiai, H., & Esaki, H. (2020). Scan-Based for Cyberattack Detection in IoMT. IEEE
Self Anomaly Detection: Client-Side Mitigation Transactions on Industrial Informatics, pp. 1-9.
of Channel-Based Man-in-the-Middle Attacks https://doi.org/10.1109/TII.2022.3231424.
Against Wi-Fi. In 2020 IEEE 44th Annual Khurana, D., Koli, A., & Khatter, K. (2023). Natural
Computers, Software, and Applications language processing: state of the art, current
Conference (COMPSAC), pp. 1498-1503. trends and challenges. Multimed. Tools Appl., 82,
https://doi.org/10.1109/COMPSAC48688.2020.0 3713–3744.
0-43. https://doi.org/10.1007/s11042-022-13428-4
Gopinath, N., & Shyry, S. P. (2023). Side Channel Attack Kreuzberger, D., Kühl, N., & Hirschl, S. (2023). Machine
Free Quantum Key Distribution Using Entangled learning operations (mlops): Overview,
Fuzzy Logic. Braz. J. Phys., 53, 35. definition, and architecture. IEEE Access, 11,
https://doi.org/10.1007/s13538-022-01246-w 31866 -31879.
GSR, E. S., Ganeshan, R., Jingle, I. D. J., & Ananth, J. P. https://doi.org/10.1109/ACCESS.2023.3262138
(2023). FACVO-DNFN: Deep learning-based Kurani, A., Doshi, P., & Vakharia, A. (2023). A
feature fusion and Distributed Denial of Service Comprehensive Comparative Study of Artificial
attack detection in cloud computing. Knowledge- Neural Network (ANN) and Support Vector
Based Systems, 261, 110132. Machines (SVM) on Stock Forecasting. Ann.
https://doi.org/10.1016/j.knosys.2022.110132 Data. Sci., 10, 183–208.
Ha, G., Chen, H., Jia, C., & Li, M. (2022). Threat model https://doi.org/10.1007/s40745-021-00344-x
and defense scheme for side-channel attacks in Kwekha-Rashid, A.S., Abduljabbar, H.N., & Alhayani,
client-side deduplication. Tsinghua Science and B. (2023). Coronavirus disease (COVID-19)
Technology, 28(1), 1-12. cases analysis using machine-learning
https://doi.org/10.26599/TST.2021.9010071. applications. Appl. Nanosci., 13, 2013–2025.
Iban, M. C., & Bilgilioglu, S.S. (2023). Snow avalanche https://doi.org/10.1007/s13204-021-01868-7
susceptibility mapping using novel tree-based Lu, Y., Qi, Y., Qi, S., Zhang, F., Wei, W., Yang, X., &
machine learning algorithms (XGBoost, Dong, X. (2021). Secure deduplication-based
NGBoost, and LightGBM) with eXplainable storage systems with resistance to side-channel
Artificial Intelligence (XAI) approach. Stochastic attacks via fog computing. IEEE Sensors
Environmental Research and Risk Journal, 22(18), 17529-17541.
Assessment, 37(6), 2243-2270. https://doi.org/10.1109/JSEN.2021.3052782.
https://doi.org/10.1007/s00477-023-02392-6 Ma, T., Xu, C., Yang, S., Huang, Y., an, Q., Kuang, X.,
Jain, A., & Rajak, R. (2023). A systematic review of & Grieco, L. A. (2023). A Mutation-Enabled
workflow scheduling techniques in a fog Proactive Defense against Service-Oriented Man-
environment. International Journal of in-The-Middle Attack in Kubernetes. IEEE
Experimental Research and Review, 30, 100-108. Transactions on Computers, pp. 1-14.
https://doi.org/10.52756/ijerr.2023.v30.011 https://doi.org/10.1109/TC.2023.3238125
Joshi, A., Capezza, S., Alhaji, A., & Chow, M. Y. (2023). Mohy-eddine, M., Guezzaz, A., Benkirane, S., & Azrour,
Survey on AI and Machine Learning Techniques M. (2023). An efficient network intrusion
for Microgrid Energy Management detection model for IoT security using K-NN
Systems. IEEE/CAA Journal of Automatica classifier and feature selection. Multimedia Tools
Sinica, 10(7), 1513-1529. and Applications, pp. 1-19.
DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
82
 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)
https://doi.org/10.1007/s11042-023-14795-2 Learning Strategies Using the Naïve Bayes
Paramanik, A. R., Sarkar, S., & Sarkar, B. (2022). Classifier Algorithm. Sinkron: Jurnal dan
OSWMI: An objective-subjective weighted Penelitian Teknik Informatika, 8(1), 256-267.
method for minimizing inconsistency in multi- https://doi.org/10.33395/sinkron.v8i1.11954
criteria decision making. Computers & Industrial Sarkar, S., Pramanik, A., Maiti, J., & Reniers, G. (2020).
Engineering, 169, 108138. Predicting and analyzing injury severity: A
https://doi.org/10.1016/j.cie.2022.108138 machine learning-based approach using class-
Patel, S. K. (2022). Attack detection and mitigation imbalanced proactive and reactive data. Safety
scheme through novel authentication model Science, 125, 104616.
enabled optimized neural network in smart https://doi.org/10.1016/j.ssci.2020.104616
healthcare. Computer Methods in Biomechanics Sarkar, S., Vinay, S., Djeddi, C., & Maiti, J. (2021). Text
and Biomedical Engineering, pp. 1-27. mining-based association rule mining for incident
https://doi.org/10.1080/10255842.2022.2045585 analysis: a case study of a steel plant in India.
Patel, S.K. (2022). Attack detection and mitigation In Pattern Recognition and Artificial Intelligence:
scheme through novel authentication model 4th Mediterranean Conference, MedPRAI 2020,
enabled optimized neural network in smart Hammamet, Tunisia, December 20–22, 2020,
healthcare. Computer Methods in Biomechanics Proceedings Springer International Publishing,
and Biomedical Engineering, pp. 1-27. 4, 257-273. https://doi.org/10.1007/978-3-030-
https://doi.org/10.1080/10255842.2022.2045585 71804-6_19
Pramanik, A., Sarkar, S., & Maiti, J. (2021). A real-time Sarkar, S., Vinay, S., Raj, R., Maiti, J., & Mitra, P.
video surveillance system for traffic pre-events (2019). Application of optimized machine
detection. Accident Analysis & Prevention, 154, learning techniques for prediction of occupational
106019. accidents. Computers & Operations
https://doi.org/10.1016/j.aap.2021.106019 Research, 106, 210-224.
Radhakishan, V., & Selvakumar, S. (2011, September). https://doi.org/10.1016/j.cor.2018.02.021
Prevention of man-in-the-middle attacks using ID Sultan, A. B. M., Mehmood, S., & Zahid, H. (2022). Man
based signatures. IEEE, In 2011 Second in the Middle Attack Detection for MQTT based
International Conference on Networking and IoT devices using different Machine Learning
Distributed Computing, 165-169. Algorithms. IEEE, in 2022 2nd International
https://doi.org/10.1109/ICNDC.2011.40 Conference on Artificial Intelligence (ICAI), pp.
Rajak, R., Choudhary, A., & Sajid, M. (2023). Load 118-121.
balancing techniques in cloud platform: A https://doi.org/10.1109/ICCCNT54827.2022.998
systematic study. International Journal of 4365.
Experimental Research and Review, 30, 15-24. Utukuru, S., Pisipati, R. K., & Karlapalem, K. (2023).
https://doi.org/10.52756/ijerr.2023.v30.002 Missing Data Resilient Ensemble Subspace
Ren, M., Tian, Y., Kong, S., Zhou, D., & Li, D. (2020, Decision Tree Classifier. In Proceedings of the
June). An detection algorithm for ARP man-in- 6th Joint International Conference on Data
the-middle attack based on data packet Science & Management of Data (10th ACM
forwarding behavior characteristics. In 2020 IKDD CODS and 28th COMAD), pp. 104-107.
IEEE 5th Information Technology and https://doi.org/10.1145/3570991.3571006
Mechatronics Engineering Conference (ITOEC), Verma, R., & Chandra, S. (2023). RepuTE: A soft voting
IEEE, 1599-1604. ensemble learning framework for reputation-
https://doi.org/10.1109/ITOEC49072.2020.9141555. based attack detection in fog-IoT milieu.
Sahi, A., Lai, D., Li, Y., & Diykh, M. (2017). An Engineering Applications of Artificial
efficient DDoS TCP flood attack detection and Intelligence, 118, 105670.
prevention system in a cloud environment. IEEE https://doi.org/10.1016/j.engappai.2022.105670
Access, 5, 6036-6048. Wang, N., Guo, H., Jing, Y., Zhang, Y., Sun, B., Pan, X.,
https://doi.org/10.1109/ACCESS.2017.2688460. Chen, H., Xu, J., Wang, M., Chen, Xi, Song, L.,
Saleh, A., Dharshinni, N. P., Perangin-Angin, D., Azmi, & Cui, W. (2023). Development and validation
F., & Sarif, M. I. (2023). Implementation of of risk prediction models for large for gestational
Recommendation Systems in Determining age infants using logistic regression and two
DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
83
 Int. J. Exp. Res. Rev., Special Vol. 31: 74-84 (2023)
machine learning algorithms. Journal of https://doi.org/10.1109/COMST.2015.2487361.
Diabetes, 15(4), 338-348. Yu, J., Yin, H., Xia, X., Chen, T., Li, J., & Huang, Z.
https://doi.org/10.1111/1753-0407.13375 (2023). Self-supervised learning for
Wu, K., Xu, Z., Lyu, X., & Ren, P. (2023). Cross- recommender systems: A survey. IEEE
supervised learning for cloud Transactions on Knowledge and Data
detection. GIScience & Remote Sensing, 60(1), Engineering,
2147298. https://doi.org/10.1109/TKDE.2023.3282907.
https://doi.org/10.1080/15481603.2022.2147298 Zhang, Y., Mao, Y., Xu, M., Xu, F., & Zhong, S. (2019).
Yan, Q., Yu, F. R., Gong, Q., & Li, J. (2015). Software- Towards thwarting template side-channel attacks in
defined networking (SDN) and distributed denial secure cloud deduplications. IEEE Transactions on
of service (DDoS) attacks in cloud computing Dependable and Secure Computing, 18(3), 1008-
environments: A survey, some research issues, 1018. https://doi.org/10.1109/TDSC.2019.2911502.
and challenges. IEEE Communications Surveys
& Tutorials, 18(1), 602-622.

How to cite this Article:

Animesh Kumar, Sandip Dutta and Prashant Pranav (2023). Supervised learning for Attack Detection in Cloud. International Journal of
Experimental Research and Review, 31, 74-84.
DOI : https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008

DOI: https://doi.org/10.52756/10.52756/ijerr.2023.v31spl.008
84