Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964519

Study on Real-Time Data Integrity and Security

Module Using Machine Learning
Greeva Dinesh Kumar Patel1; Kartikeydheer Srivastava2; Mansi Mehta3
1
UG Student, Department of CSE, Indus Institute of Technology and Engineering (IITE),
Indus University, Ahmedabad, Gujarat, India
2
UG Student, Department of CSE, Indus Institute of Technology and Engineering (IITE),
Indus University, Ahmedabad, Gujarat, India
3
Assistant Professor, Department of CSE, Indus Institute of Information & Communication Technology (IIICT),
Indus University, Ahmedabad, Gujarat, India

Publication Date: 2025/03/06

Abstract: The rapid increase of interconnected virtual ecosystems, pushed by means of IoT devices, cloud infrastructures,
and software program-described Networks (SDNs), has amplified the call for for sturdy actual-time information safety and
integrity. This studies introduces a real-Time data Integrity and security Module (RTDISM) that leverages system getting
to know (ML) to tackle these challenges. by way of integrating SDN architectures, light-weight federated getting to know,
and adaptive authentication protocols, the RTDISM provides a scalable, multi-layered defense gadget suitable for diverse
environments. Key capabilities include adaptive ML-pushed risk mitigation, low-latency responses, and efficient aid usage.
The module is particularly proper for vital domain names together with healthcare, commercial automation, and self reliant
systems. Innovat ions include SDN-primarily based security for improved network flexibility, light-weight ML models for
aid optimization, and federated getting to know for decentralized, privateness-maintaining operations. Experimental
opinions demonstrate the RTDISM's superior overall performance in accuracy, reaction time, and useful resource
performance in comparison to present solutions, organising it as a benchmark for subsequent- technology cybersecurity
structures.

Keywords: Real-Time Security, ML, IoT, SDN, Anomaly Detection, Edge AI, Federated Learning, Threat Mitigation.

How to Cite: Greeva Dinesh Kumar Patel; Kartikeydheer Srivastava; Mansi Mehta (2025). Study on Real-Time Data Integrity
and Security Module Using Machine Learning. International Journal of Innovative Science and Research Technology,
10(3), 239-245. https://doi.org/10.5281/zenodo.14964519

I. INTRODUCTION financial systems[1][2]. IoT gadgets, projected to surpass

seventy five billion via 2025, call for sturdy safety answers to
The exponential boom of virtual ecosystems, pushed combat threats like DDoS assaults, ransomware, and insider
with the aid of IoT devices, cloud infrastructures, and breaches[1]. traditional static safety frameworks fail to
software program-described Networks (SDNs), has address these dynamic threats, necessitating adaptive, clever
revolutionized information processing while introducing vital techniques.
vulnerabilities in healthcare, commercial automation, and

Fig 1 Cybersecurity Threats and Mitigation Techniques (2020-2024)

Machine learning (ML) offers transformative solutions and Convolutional Neural Networks (CNNs)[1][4]. This look
by using autonomously reading datasets, detecting anomalies, at introduces a real-Time information Integrity and security
and predicting vulnerabilities with strategies like assist Module (RTDISM), integrating ML algorithms, federated
Vector Machines (SVM), decision trees (DT), Autoencoders, learning, and edge AI for scalable, adaptive cybersecurity.

IJISRT25MAR069 www.ijisrt.com 1
Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964519
RTDISM minimizes latency and complements privateness The RTDISM leverages SDN architectures for
through processing records regionally on IoT gadgets and centralized manage and dynamic chance mitigation[1][3]. It
part servers, decreasing reliance on centralized cloud additionally carries light-weight cryptographic strategies and
structures[15]. adaptive authentication mechanisms to guard touchy IoMT-
generated health statistics towards man-in-the-middle
(MITM) assaults, replay attacks, and impersonation[11].
Empirical evaluations highlight its superior performance in
accuracy, response time, and resource efficiency.

Fig 2 Resource Utilization Breakdown

Fig 3 Publication Date Breakdown- Authentication Protocols

By synthesizing advancements in IoT, SDN, and cloud gadgets lack the computational electricity for traditional
security, RTDISM sets a benchmark for next-generation cryptographic algorithms, and inconsistent standards
cybersecurity frameworks, addressing the critical need for amongst providers create vulnerabilities[5].system studying
real-time data integrity in an interconnected digital offers dynamic, adaptive processes to hazard detection, with
ecosystem. supervised techniques like guide Vector Machines (SVM)
and Random forest excelling in established statistics class, at
II. SURVEYS ARTICLES FOR THE IOT the same time as unsupervised strategies including clustering
and Autoencoders efficaciously detectanomalies in
IoT security faces precise demanding situations due to unstructured data. These approaches address evolving
device heterogeneity, resource constraints, and fragmented threats by analyzing behavioral patterns and data anomalies
ecosystems, requiring tailor-made answers. Many IoT in real time.

IJISRT25MAR069 www.ijisrt.com 2
Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964519
Deep learning has advanced cybersecurity by leveraging
Convolutional Neural Networks (CNNs) and Recurrent
Neural Networks (RNNs) to analyze full-size datasets,
discover latent malicious styles, and enhance hazard
detection talents. these strategies drastically enhance
Intrusion Detection structures (IDS) and malware category
frameworks via accomplishing higher detection costs and
minimizing fake positives. This integration creates a multi-
faceted safety system that successfully addresses the
complexity of evolving cyber threats. IoMT systems face
demanding situations in securing statistics transmission and
garage. [5]proven SDN's capability to decorate community
control and safety, integrating device learning to optimize
resource utilization, statistics shipping, and real-time routing
even as decreasing verbal exchange overhead[5][8][12].

Fig 4 Areas of Research for IOT Survey Articles

Fig 5 Historical Threats Trends and Research Publication

Research in IoT protection makes a speciality of this examine proposes a strong protection framework utilising
addressing demanding situations like code injection, IoMT ML algorithms and formal verification to make sure data
vulnerabilities, and authentication[6]proposed CNN-based integrity and security throughout present day IoT
fashions for code injection detection, the usage of character- environments.
degree n-grams to perceive malicious code patterns. [16]
highlighted SDN's position in IoMT structures, combining III. THREAT MODELS
device studying to optimize actual-time statistics routing,
expect useful resource usage, and reduce communication IoT threat fashions cope with vulnerabilities arising
overhead. [18] reviewed over 40 lightweight authentication from interconnected, useful resource-confined devices and
protocols, emphasizing relaxed methods like ECC for M2M insufficient security protocols. commonplace threats include
communication and protection towards attacks inclusive of network eavesdropping, DDoS attacks, and device spoofing,
guy-in-the-middle and impersonation. which take advantage of the heterogeneity of IoT ecosystems.
Attackers often use compromised gadgets to infiltrate
Anomaly detection frameworks leverage statistical networks, developing cascading vulnerabilities. The reliance
models and superior ML techniques for real-time hazard on wi-fi conversation protocols similarly increases risks, as
identity. Blockchain era has been explored for enhancing those channels are susceptible to interception and
information integrity and decentralized authentication, manipulation. advanced system studying algorithms, mainly
though scalability challenges persist. A habitual theme is the anomaly detection fashions, play a critical position in figuring
want for standardized taxonomies to unify security measures out and mitigating these threats in real- time, enabling
across numerous IoT ecosystems.building on those insights, proactive and adaptive security features[5][16][18].

IJISRT25MAR069 www.ijisrt.com 3
Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964519
Table 1 Summary of Threats and Mitigation Techniques
Year Threats Identified Types of Threats Mitigation Techniques
2024 320 IoT Botnets, Data Exfiltration Anomaly detection using ML, predictive analytics, and
endpoint security orchestration[14][15].
2023 270 Ransomware 2.0, Cloud Malware AI-driven behavioral detection, cloud security protocols,
Injections and automated incident mitigation[15][17].
2022 230 Zero-Day Exploits, Cross- Site Scripting Proactive vulnerability scanning, sandboxing, and
(XSS) Attacks regular system patching[16][18].
2021 200 Data Breaches, Advanced Persistent Encryption algorithms, realtime threat monitoring, and
Threats (APTs) behavioral analytics[15][16].
2020 160 Supply Chain Attacks, Cloud Blockchain verification systems, AI-based
Misconfigurations misconfiguration analysis
2019 130 Malware, IoT Exploitation, Insider Secure firmware updates, device authentication, and
Threats privilege-based access control[16][18].
2018 100 DDoS, Phishing, Ransomware Traffic filtering, multi-factor authentication, and robust
incident response plans[18][19].

Lightweight authentication protocols, such as Elliptic accuracy. Autoencoders have shown promise in identifying
Curve Cryptography (ECC) and hash-based totally methods, network intrusions by learning baseline behaviors and
provide sturdy security with minimum computational flagging deviations. Statistical models are effective for
overhead, making them ideal for IoT environments[1]. ECC structured data but are increasingly complemented by ML
supports secure device-to-tool communication, even as techniques to address dynamic and unstructured IoT
symmetric key protocols and public-key infrastructures offer environments.
alternatives with alternate-offs in safety and aid performance.
Multi-thing authentication, even though useful resource- Performance evaluations of the Data Integrity and
intensive, complements safety against identity spoofing and Security Module emphasize its effectiveness across key
unauthorized get right of entry to. real-time anomaly metrics: detection accuracy, prediction accuracy, and
detection is critical for IoT security. [4] and others spotlight response latency. These metrics demonstrate its capability to
device learning methods, which include supervised models address diverse threat models and safeguard IoT ecosystems
using labeled datasets, unsupervised models detecting in real-time.
deviations, and hybrid models combining both for improved

Fig 6 Accuracy and Latency Metrics

The module achieves an great detection accuracy of a prediction accuracy of 98.1%, the module successfully
99.9%, showcasing its capacity to identify anomalies and forecasts rising threats by using studying historic information
complex threats, which include guy-in-the- middle (MITM) and behavioral patterns. This predictive capability is essential
and Replay attacks, in actual-time. This excessive precision for addressing advanced assault vectors like Sybil and
ensures reliable differentiation between everyday operations Impersonation assaults. The module’s proactive danger
and capacity safety breaches, minimizing fake positives and anticipation permits the implementation of mitigation
enhancing consider within the system's detection talents.With strategies earlier than attacks materialize, strengthening the

IJISRT25MAR069 www.ijisrt.com 4
Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964519
general protection framework[18]. A reaction latency of one Lightweight cryptographic techniques are essential for
hundred twenty milliseconds highlights the module’s rapid securing data transmission and storage, particularly in
reaction to recognized threats, a critical feature for actual- resource-constrained IoT environments. Elliptic Curve
time IoT systems. This low latency is particularly massive for Cryptography (ECC) is a standout solution, offering robust
mitigating time- touchy assaults along with MITM, ensuring encryption with minimal computational demands, making it
ongoing communications remain comfy and information suitable for IoT devices with limited processing power[3].
integrity is preserved. Symmetric key protocols also play a vital role, providing fast
and efficient encryption by utilizing shared secret keys.
These metrics validate the module's robust performance Advanced dynamic protocols, incorporating timestamps,
in safeguarding IoT ecosystems. The combination of near- nonces, and hash functions, further bolster security by
perfect detection accuracy, predictive precision, and swift preventing common threats like replay attacks and ensuring
response times ensures comprehensive protection against the integrity of communication sessions. These combined
diverse and evolving threat models, positioning the module measures create a layered defense framework, addressing the
as a reliable and adaptive solution for IoT unique challenges of IoT security effectively[17][18].
security[17][18][19].
Dynamic authentication protocols are vital for securing
IV. COUNTERMEASURES AND FORMAL IoT communications with the aid of leveraging techniques
SECURITY VERIFICATION like timestamps, random numbers (nonces), and hash
features. Timestamps ensure that transactions are clean,
Addressing IoT safety demanding situations requires a stopping replay assaults with the aid of stopping the reuse of
multi-faceted technique that mixes superior technical answers intercepted records packets. Nonces introduce randomness to
with formal verification techniques to create a robust make each session particular, and hash features securely
protection against evolving threats. One important degree is rework facts into fixed-size strings, that are tough to opposite-
the deployment of gadget mastering (ML)-powered Intrusion engineer or mirror. these protocols are particularly powerful
Detection structures (IDS), which excel in figuring out and in useful resource-constrained IoT environments, providing a
mitigating cyber threats in actual time. Autoencoders, for lightweight but strong option to cozy communique.
example, examine baseline network behaviors and flag
deviations which can indicate anomalies or ability attacks. Software-Defined Networking (SDN) enhances IoT
further, ensemble models, which integrate a couple of security by centralizing network management, allowing for
algorithms like Random Forests and help Vector Machines dynamic reconfiguration of the network. SDN separates the
(SVMs), improve hazard detection accuracy by way of control plane from the data plane, enabling quick detection
leveraging their blended strengths. those systems are further and isolation of compromised devices, preventing attacks
more advantageous by adaptive mastering capabilities, letting from spreading. The programmability of SDN supports the
them evolve constantly by incorporating new information, integration of ML algorithms to analyze network traffic
making sure they continue to be effective against rising attack patterns, predict potential security breaches, and optimize
vectors[4][16]. resource allocation, ensuring both security and efficiency in
IoT networks.

Fig 7 Main Features and Challenges of the Framework

IJISRT25MAR069 www.ijisrt.com 5
Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964519
The chart illustrates the twin nature of IoT security, statistics breaches. Mitigation techniques consisting of
wherein excessive function significance highlights the comfortable boot mechanisms for devices and encryption
framework's strengths, while top notch venture importance protocols for networks are hired. Standardization efforts
points to areas requiring ongoing refinement. This balance purpose to address interoperability gaps, allowing seamless
emphasizes the evolving nature of developing sturdy IoT integration at the same time as ensuring compliance with
security answers, in which strengths should be continuously safety policies. however, the absence of universally ordinary
stepped forward to meet emerging demanding situations. the standards remains a assignment, highlighting the want for
integration of blockchain generation has received sizable industry collaboration [9][10].
interest for boosting IoT protection. Blockchain’s
decentralized, immutable ledger guarantees statistics Authentication protocols play a crucial role in
integrity and stops tampering across dispensed networks, establishing trust within IoT ecosystems. These protocols are
making sure duty. [2][7]emphasize its role in facilitating at categorized based on their cryptographic mechanisms,
ease, obvious interactions between IoT devices. lightweight resource efficiency, and scalability. Lightweight methods,
consensus mechanisms, including proof of Authority (PoA) such as Elliptic Curve Cryptography (ECC), are suited for
and Delegated evidence of Stake (DPoS), deal with resource-constrained devices, whereas more robust methods
scalability issues in huge networks. moreover, smart contracts like RSA provide higher security levels for critical
automate safety protocols, decreasing the need for manual applications.
interventions in risk mitigation
The comparative analysis highlights the trade-offs
 Taxonomy and Comparison of Authentication Protocols between computational overhead and security effectiveness
for the IoT in IoT authentication protocols. Machine learning (ML)
This phase delves into the authentication protocols models enhance these protocols by detecting anomalies and
utilized in actual-time statistics integrity and protection unauthorized access without significantly increasing resource
fashions for IoT structures. A cohesive taxonomy of threats, demands. Various authentication protocols are designed for
countermeasures, and requirements is important to advancing specific subdomains, addressing their unique security
IoT protection. existing literature categorizes IoT threats into requirements. This section presents a taxonomy of these
three primary layers: device, network, and alertness. device- protocols, comparing their effectiveness in securing different
stage threats regularly involve firmware tampering, whilst IoT environments.
community-stage threats encompass eavesdropping and

Fig 8 Performance Metrics of Real-Time Data Integrity and Security Module

The class file is designed for, but not limited to, six M2M conversation permits direct device interaction in
authors. A minimum of one author is required for all IoT, emphasizing relaxed and green conversation in spite of
conference articles. Author names should be listed starting resource constraints. light-weight cryptographic strategies
from left to right and then moving down to the next line. This like ECC and hash-primarily based protocols are usually
is the author sequence that will be used in future citations and used, imparting strong safety while minimizing
by indexing services. Names should not be listed in columns computational overhead, critical for dynamic and disbursed
nor grouped by affiliation. Please keep your affiliations as networks.
succinct as possible (for example, do not differentiate among
departments of the same organization) [15][16][18].

IJISRT25MAR069 www.ijisrt.com 6
Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964519
IoV connects vehicles, infrastructure, and entities to quantum-resistant authentication mechanisms. artificial
improve road safety and traffic management. Authentication intelligence is increasingly being integrated into
protocols in IoV focus on secure data transmission and authentication protocols, supplying adaptive security features
integrity. PKI, digital certificates, and blockchain technology and real-time hazard detection through anomaly detection and
are key for establishing secure communication and ensuring behavioral analysis.
data immutability and transparency.
Future research should focus on developing advanced
IoE integrates IoT in smart grids and energy authentication frameworks that address the evolving
management. Authentication protocols must support real- challenges of IoT security. Interdisciplinary collaboration
time data exchange and secure communication across devices among cybersecurity researchers, cryptographers, and AI
like smart meters and control systems. These protocols are specialists is essential to designing authentication solutions
designed to protect against cyber threats, utilizing time- that balance security, computational efficiency, and
sensitive mechanisms and strong encryption to safeguard scalability. Establishing universally accepted authentication
energy infrastructures. standards will play a crucial role in ensuring interoperability
across heterogeneous IoT ecosystems. Research into privacy-
IoS involves resource-constrained sensors used in preserving authentication mechanisms, such as homomorphic
environments like healthcare and environmental monitoring. encryption and secure multi-party computation, can further
Authentication protocols focus on minimal resource use enhance security without compromising user privacy.
while maintaining security. Lightweight techniques, such as
symmetric key cryptography and one-time passwords, are By means of integrating cryptographic innovations,
used, alongside machine learning models to enhance real- artificial intelligence, and scalable authentication
time threat detection and mitigation. architectures, future authentication protocols will offer sturdy
and adaptive protection solutions for the dynamic nature of
Unique IoT subdomains have unique authentication IoT environments. Addressing the safety concerns of IoT
necessities primarily based on their operational ecosystems requires a proactive approach in developing
environments. gadget-to-system (M2M) communication authentication frameworks which could face up to emerging
requires cozy and green authentication mechanisms to cyber threats at the same time as making sure seamless device
support self reliant device interactions in dispensed networks. interactions [13][16][18].
lightweight cryptographic techniques like ECC and hash-
primarily based protocols are generally employed to make V. CONCLUSION
certain safety with out excessive computational overhead.
The internet of automobiles (IoV) demands sturdy This research presents a Real-Time Data Integrity and
authentication protocols to allow secure records transmission Security Module designed to tackle the challenges of securing
among motors, infrastructure, and street entities. PKI, digital data in dynamic IoT environments. By integrating advanced
certificates, and blockchain-primarily based authentication machine learning (ML) algorithms, the framework
play a extensive position in ensuring transparency and effectively mitigates security threats across the device,
statistics integrity. network, and application layers, ensuring robust data
protection in real-time. The use of Convolutional Neural
In the Internet of Energy (IoE), authentication protocols Networks (CNNs) for anomaly detection, Software-Defined
are designed to secure data exchanges in smart grids and Networking (SDN) for adaptive control, and advanced
energy management systems. Strong encryption techniques authentication protocols together create a scalable solution
and real-time authentication mechanisms protect against capable of addressing evolving cybersecurity threats.
cyber threats while ensuring operational reliability. The
Internet of Sensors (IoS) focuses on security solutions for This demonstrates that ML-pushed security
resource-constrained environments, such as healthcare and mechanisms, mainly anomaly detection models, notably
environmental monitoring systems. Authentication enhance the accuracy and efficiency of danger identification,
techniques in these domains prioritize low-power mitigation, and response. Federated learning permits
consumption, integrating lightweight cryptographic decentralized safety operations, lowering reliance on
protocols, one-time passwords, and AI-driven threat centralized cloud infrastructures at the same time as
detection models. maintaining data privateness. moreover, SDN-based
protection solutions enhance network adaptability, allowing
IoT authentication is shaped by rising technologies that for dynamic policy enforcement and actual-time threat
enhance safety and flexibility. Federated gaining knowledge intelligence. The RTDISM exhibits tremendous overall
of is being explored to allow decentralized authentication, performance in detection accuracy, reaction latency, and
reducing reliance on centralized identification control useful resource optimization, making it well-acceptable for
structures. The 0-believe architecture is gaining traction as a essential applications which includes healthcare, industrial
paradigm that assumes no inherent consider within a automation, and independent structures.
community, requiring continuous authentication verification
at each interaction. With the advent of quantum computing, A comprehensive analysis of authentication protocols
traditional cryptographic techniques are at danger, highlights the need of standardized protection frameworks
necessitating research into submit-quantum cryptography for tailor-made to aid-constrained IoT environments. The take a

IJISRT25MAR069 www.ijisrt.com 7
Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964519
look at underscores the significance of submit-quantum Sep. 2023. [Online]. Available:
cryptography, zero-consider architectures, and AI-pushed https://doi.org/10.3390/electronics12183958
safety automation in future cybersecurity techniques. As IoT [7]. S. K. Sahu and K. Mazumdar, "Exploring Security
ecosystems preserve to expand, interdisciplinary Threats and Solutions Techniques for Internet of
collaboration among AI researchers, cybersecurity Things (IoT): From Vulnerabilities to Vigilance,"
specialists, and enterprise leaders may be important in Frontiers in Artificial Intelligence, vol. 7, no. 1397480,
growing scalable, sensible safety answers. May 2024. [Online]. Available:
https://doi.org/10.3389/frai.2024.1397480
By synthesizing improvements in ML, blockchain [8]. S. B. Masud, M. M. Rana, H. J. Sohag, F. Shikder, M.
protection, and adaptive authentication, this studies R. Faraji, and M. M. Hasan, "Understanding the
establishes the RTDISM as a benchmark for subsequent- Financial Transaction Security Through Blockchain
technology IoT protection frameworks. The version's and Machine Learning for Fraud Detection in Data
scalability, real-time adaptability, and precision-driven Privacy and Security," Pakistan Journal of Life and
security features ensure that it offers a resilient foundation for Social Sciences, vol. 22, no. 2, pp. 17782-17803, Dec.
defensive interconnected systems, retaining information 2024. [Online]. Available:
integrity, and proactively countering sophisticated cyber http://dx.doi.org/10.57239/PJLSS-2024-22.2.001296
threats in an increasingly more virtual landscape. [9]. S. K. Devineni, S. Kathiriya, and A. Shende, "Machine
Learning-Powered Anomaly Detection: Enhancing
This study emphasizes the transformative impact of Data Security and Integrity," Journal of Artificial
machine learning in modern cybersecurity, demonstrating its Intelligence & Cloud Computing, vol. 2, no. 2, pp. 1-
potential to safeguard data integrity and security in 9, May 2023. [Online]. Available:
increasingly dynamic environments. As the digital world http://dx.doi.org/10.47363/JAICC/2023(2)184
evolves, ongoing research and innovation will be crucial in [10]. J. Bhayo, S. A. Shah, S. Hameed, A. Ahmed, J. Nasir,
maintaining the security of critical data systems. and D. Draheim, "Towards a machine learning-based
framework for DDOS attack detection in software-
ACKNOWLEDGMENT defined IoT (SD-IoT) networks," Engineering
Applications of Artificial Intelligence, vol. 123, p.
The authors declare that they have no conflicts of 106432, May 2023. [Online]. Available:
interest. https://doi.org/10.1016/j.engappai.2023.106432
[11]. F. Alwahedi, A. Aldhaheri, M. A. Ferrag, A. Battah,
REFERENCES and N. Tihanyi, "Machine learning techniques for IoT
security: Current research and future vision with
[1]. M. Asmar and A. Tuqan, "Integrating machine generative AI and large language models," Internet of
learning for sustaining cybersecurity in digital banks," Things and Cyber-Physical Systems, vol. 4, pp. 167–
Heliyon, vol. 10, p. e37571, 2024. Available: 185, Jan. 2024. [Online]. Available:
https://doi.org/10.1016/j.heliyon.2024.e37571 https://doi.org/10.1016/j.iotcps.2023.12.003
[2]. K. Masthan, M. Shabana, and M. Rafi, "Enhancing [12]. H. El-Sofany, S. A. El-Seoud, O. H. Karam, and B.
cloud security using machine learning," Journal of Bouallegue, "Using machine learning algorithms to
Systems Engineering and Electronics, vol. 34, no. 10, enhance IoT system security," Scientific Reports, vol.
pp. 66–70, 2024. Available: 14, no. 12077, 2024. [Online]. Available:
https://www.researchgate.net/publication/386872942 http://dx.doi.org/10.1038/s41598-024-62861-y
[3]. M. Austin, K. Austin, and M. Osaka, "Machine [13]. M. Salayma, "Risk and threat mitigation techniques in
learning for data security in cloud computing internet of things (IoT) environments: A survey,"
environments," Research Proposal, 2024. DOI: Frontiers in the Internet of Things, vol. 2, no.
10.13140/RG.2.2.12840.14088. Available: 1306018, Jan. 2024. [Online]. Available:
http://dx.doi.org/10.13140/RG.2.2.12840.14088 https://doi.org/10.3389/friot.2023.1306018
[4]. D. Alexander and Z. Chain, "Data Integrity Challenges [14]. L. Doris and R. Shad, "Using machine learning
and Solutions in Machine Learning-driven Clinical models to identify and predict security-related
Trials," Sher-e-Kashmir University of Agricultural anomalies in real-time for proactive maintenance,"
Sciences and Technology of Kashmir, Aug. 2023. ResearchGate, Dec. 2024. Available:
[Online]. Available: https://www.researchgate.net/publication/386573125_
https://www.researchgate.net/publication/373214664 USING_MACHINE_LEARNING_MO DELS_TO_
[5]. E. Altulaihan, M. A. Almaiah, and A. Aljughaiman, IDENTIFY_AND_PREDICT_SECURITY-
"Cybersecurity Threats, Countermeasures and _RELATED_ANOMALIES_IN_REALTIME_
Mitigation Techniques on the IoT: Future Research FOR_PROACTIVE_MAINTENA NCE
Directions," Electronics, vol. 11, no. 3330, Oct. 2022. [15]. G. Arjunan, "Optimizing Edge AI for Real-Time Data
[Online]. Available: Processing in IoT Devices: Challenges and
https://doi.org/10.3390/electronics11203330 Solutions," International Journal of Scientific
[6]. T. S. AlSalem, M. A. Almaiah, and A. Lutfi, Research and Management (IJSRM), vol. 11, no. 6, pp.
"Cybersecurity Risk Analysis in the IoT: A 944-953, June 2023. [Online]. Available:
Systematic Review," Electronics, vol. 12, no. 3958, DOI:10.18535/ijsrm/v11i06.ec2

IJISRT25MAR069 www.ijisrt.com 8
Volume 10, Issue 3, March – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964519
[16]. A. S. Shaik and A. Shaik, "Code Injection Attack
Prevention with AI- Integrated Machine Learning
Approach Using CNN," ShodhKosh: Journal of
Visual and Performing Arts, vol. 3, no. 2, pp. 848-854,
Dec. 2022. [Online]. Available:
https://doi.org/10.29121/shodhkosh.v3.i2.2022.3181
[17]. M. Sugadev et al., "Implementation of Combined
Machine Learning with the Big Data Model in IoMT
Systems for the Prediction of Network Resource
Consumption and Improving the Data Delivery,"
Computational Intelligence and Neuroscience, vol.
2022, Article ID 6510934, 12 pages, July 2022.
[Online]. Available:
https://doi.org/10.1155/2022/6510934
[18]. M. A. Ferrag, L. A. Maglaras, H. Janicke, J. Jiang, and
L. Shu, "Authentication Protocols for Internet of
Things: A Comprehensive Survey," Security and
Communication Networks, vol. 2017, Article ID
6562953, pp. 1– 41, 2017, [Online]. Available:
https://doi.org/10.1155/2017/6562953
[19]. N. Singh, R. Buyya, and H. Kim, "Securing Cloud-
Based Internet of Things: Challenges and
Mitigations," Sensors, vol. 25, no. 79, pp. 1–45, 2025,
[Online]. Available:
https://doi.org/10.3390/s25010079

IJISRT25MAR069 www.ijisrt.com 9