2024 5th International Conference on Industrial Engineering and Artificial Intelligence (IEAI)

Biometrics in the age of AI: Comprehensive study on

2024 5th International Conference on Industrial Engineering and Artificial Intelligence (IEAI) | 979-8-3503-8636-3/24/$31.00 ©2024 IEEE | DOI: 10.1109/IEAI62569.2024.00016

the usage of biometrics in the Philippines against

falsification through artificial intelligence
Eric Blancaflor Arjun Marcus Bali Matthew Allister Kitan Jewl Danielle Magbanue
School of Information School of Information School of Information School of Information
Technology Technology Technology Technology
Mapua University Mapua University Mapua University Mapua University
Makati, Philippines Makati, Philippines Makati, Philippines Makati, Philippines
[email protected] [email protected] [email protected] [email protected]
u.ph

Abstract— Integration of biometric technology for personal manipulating existing footage to depict events that never
identification is of growing interest and also an area of concern. occurred or words never spoken. This juxtaposition emphasizes
Nevertheless, emerging technological integration emerges as a the necessity for a nuanced exploration of the integration of
subject of interest in academic interest and the evolving global biometrics and the potential risks posed by emerging
market, with the Philippines being no exception. The growing technologies like deepfakes in the realm of technological
misuse and development of AI remain a threat with regard to security.
falsifying biometric data. While such technology is potentially
destructive to an entity, embracing how tech will continue to While researchers are still exploring the frontiers of
evolve while preparing and mitigating the risks is ideal to thrive artificial intelligence (AI) and continue to advance in creating
as an institution. This comprehensive study aims to learn about innovative technologies, there is still a need to examine the
the current methods of biometrics common in the Philippines, possible risks or vulnerabilities surrounding biometric
namely fingerprint and facial ID in areas such as banking and integration. This paper delves into the risks of integrating
the national ID system, highlighting key vulnerabilities to AI biometrics and AI, and ascertain concerns that possibly lead to
methods such as spoofing and social engineering through the use privacy issues. The existence and accessible use of deepfakes
of falsified images and possible solutions to mitigate falsification. which may manipulate biometric data and how threats emerge,
casts uncertainty on biometric security measures. Thus why
Keywords— AI, Biometrics, Deepfakes, Neural Network through this paper, it is crucial to weigh in between the
nuanced vulnerabilities and benefits this integration may serve
I. INTRODUCTION comprehensively.
A. Background of the Study B. Objective of the study
Biometric integration currently stands as another one of The paper serves as a comprehensive resource, providing
mankind's milestones in technological advancement. The use essential information regarding biometrics, particularly AI, and
of distinctive biological and behavioral features is promised to the associated threats. It endeavors to highlight potential and
provide another unique, easy, and new degree of accuracy as historical risks, address how these threats manifest, and supply
an alternative to inputting sensitive credential information available research in managing these threats. Additionally, the
ranging from fingerprint recognition to face scans. Integrating paper explores possible alternative approaches to attaining
biodata into authentication and identification systems has its technological security based on available research information.
praise and risks, giving room for debate as to whether this new With this contribution, the reader may participate in creating
technology would do more good than harm as a way to informed discussions supporting strategic decision-making in
automatically input the credentials of the user. This is technological security.
especially critical when this integration would be implemented
in financial institutions where ease of access is still prone to II. LITERATURE REVIEW
compromising data security with the prevalent threats
This section compiles academic studies and related
surrounding deepfakes, for instance as another form of
literature for the researchers to consider in biometrics involving
Artificial Intelligence.
the potent threats or vulnerabilities in its integration.
Biometrics involves measuring and statistically analyzing
A. Artificial Intelligence today
people's unique physical and behavioral characteristics, such as
fingerprints, facial features, or voice patterns, for identification Biometrics and Decision Support Systems (DSS) research
and authentication. On the other hand, deepfakes represent a emphasizes the mandatory integration of biometrics in complex
technology that utilizes deep learning algorithms to create systems, including security checkpoints, health monitoring,
realistic-looking fake videos or audio recordings, often robots, and epidemiological surveillance [1]. According to Lai

979-8-3503-8636-3/24/$31.00 ©2024 IEEE 42

DOI 10.1109/IEAI62569.2024.00016
Authorized licensed use limited to: VTU Consortium. Downloaded on February 24,2025 at 18:28:41 UTC from IEEE Xplore. Restrictions apply.
et al. (2020), the introduced Risk, Trust, and Bias (R-T-B) themes emphasizes the importance of balancing technological
framework acknowledges these emerging measures' impact on advancements with ethical considerations, privacy protection,
system performance. The study proposes a comprehensive and the prevention of malicious use.
taxonomy linking R-T-B assessment to causal inference, Newer and more sophisticated methods of spoofing
showcasing practical applications in synthetic biometric trust biometrics come with the advancements of technology. They
assessment and face biometric bias risk evaluation. are a challenge among those aiming to secure themselves and
Methodological research on the legal regulation models in their organizations. [6] According to Bodepudi and Reddy, one
biometric identification adopts a risk-oriented approach to law, of the most common spoofing attacks involving AI would be
focusing on legal regulation models for biometric identification, through the usage of deepfakes that are currently on the rise
especially facial recognition technology [2]. According to with the popularity of software such as face swapping online.
Utegen and Rakhmetov (2023), the objective is to specify these Demonstrating the need for constant updating of anti-spoofing
models to enhance information security and protect individuals' models and databases as such data for attacks can easily come
privacy rights. The methods include comparative-legal analysis from something as simple as a social media post. The proposed
and juridical forecasting, anticipating risks related to the counter to such attacks, according to them, is the usage of
further dissemination of facial recognition technology [2]. algorithms such as Random Forest, Support Vector Machines,
and Convolutional Neural Networks after facial extraction.
Research about securitization in the post-9/11 era delves
into the securitization concept post-9/11, where new threats B. AI and Privacy
justify exceptional measures like increased political The paper entitled “Emotional AI, Soft Biometrics, and the
surveillance [3]. According to Hayek (2022), the study Surveillance of Emotional Life: An Unusual Consensus on
explores the relationship between privacy and security in an era Privacy,” authored by Andrew McStay, talks about issues
of uncertainty, comparing Western democracies to China's regarding privacy with the use of emotion-centered AI. [7] The
model. Highlighted is the absence of social credit systems and paper concludes that there exists a limited window of
biometric tools in the West, signaling potential risks in the opportunity to societally agree on principles of practice
digitalization journey. regarding privacy and the use of data about emotions. The
An exploratory research on Asian Americans' biometric paper closely examines the privacy issues of emotional AI
acceptance focuses on an exploratory qualitative study approaches that seem to steer clear of identifying specific
assessing the acceptance of biometric technology among Asian persons.
Americans [4]. According to Nguyen (2023), convenience The paper entitled “Biometrics and AI Bias” by K. Michael,
emerges as a key driver, with participants appreciating the ease R. Abbas, P. Jayashree, R. J. Bandara, and A. Aloudat,
of biometric authentication for accessing their phones. Despite describes biometrics as a distinct physical characteristic in
the convenience, participants expressed concerns about the denoting uniqueness [8] since the turn of the 1900s such as
security of biometric authentication, particularly for significant fingerprint classification matching in 1900 and automation of
financial accounts. The malicious use and abuse of AI this by the US law enforcement by mid-1980s. A decade later,
capabilities construct a typology of activities and 500 automatic fingerprint identification systems (AFIS) were
corresponding risks [5]. According to Blauth et al. (2022), the used to convict people of crimes, marking this the first time
study identifies vulnerabilities in AI models and explores AI- automation had been used to cross-check minutiae. There are
enabled and AI-enhanced attacks. Notably, it suggests millions of minutiae gathered worldwide using a high-
categories like integrity attacks, unintended AI outcomes, resolution camera, which traditionally originated from ink-
algorithmic trading, social engineering, misinformation, based methods and further showed how widely used even the
hacking, and autonomous weapon systems. ink-based methods were with systems like Aadhaar, for
Across the literature, there is a recurring theme of the example, with over 1,000,000,000 fingerprints collected
critical role of biometrics in diverse applications, from security systematically as the world's largest biometric ID system, and
to healthcare, with implications for trust, risk, and bias. Legal AFIS having more than 220,000 fingerprints from 17000 crime
frameworks are essential for regulating biometric technology, scene marks, enabling 3,000 comparisons a day. This shows
especially facial recognition, to address privacy concerns and how biometrics evolved, why there was a need for it, and how
information security risks. The comparison between Western it came to be, however, this was before the year 2000 was
democracies and China highlights potential cultural and mentioned, and facial biometrics was only in the last two
infrastructural foundations for advanced surveillance systems, decades when it started and became prevalent for a wide
including biometrics. While there's a general trend of variety of applications. The paper further talks about the
acceptance for biometric convenience, concerns about security technical and historical details of facial recognition systems
persist, especially in critical domains like finance. The broader and categorizes them into four stages; face detection, face
context of AI's malicious use underscores the need for analysis, image-to-data conversion using a very complex
governance strategies and policies to mitigate risks and mathematical formula, and lastly, comparison matching.
enhance resilience. There are also some challenges in this despite the
The synthesis highlights the interconnectedness of availability and advancement of high-definition cameras, since
biometrics, AI, legal regulations, and societal implications, low-resolution footage, and other camera noises (very low light,
urging a holistic approach to the responsible development and off angle, atmospheric conditions..) add to the complexity of
deployment of these technologies. The convergence of these using photographic evidence as input for this new technology

43

Authorized licensed use limited to: VTU Consortium. Downloaded on February 24,2025 at 18:28:41 UTC from IEEE Xplore. Restrictions apply.
to be able to prove the match between the guilty and the culprit.
Yet this is still a great alternative without eyewitness accounts
when the crime is committed. As of today, biometric
technology is still currently being upgraded to be more
intelligent, adding capabilities to even the lowest resolution
surveillance cameras, with web scraping images from the
Internet and social media platforms to enhance their result.
Nevertheless, another drawback of this feat arose, stating that
“the possibility of entangling innocent people in suspect lists
and even wrongful arrests are higher than ever before”,
resulting in wrongful arrests or entangling innocent people in
Figure 1 Credit: fintechnews.ph [10]
suspect lists. The benefit of this method was that the public
images provided another method to solve crimes that account In a study by Iyer et al. notes the factors to be considered
for individuals who do not yet have data on criminal databases with regard to the issues of the accuracy of biometrics, such as
and may have never come to the attention of authorities in the natural events like aging, and the flaw of twins' similar faces
first place if it wasn’t for the technology. defeating facial recognition, as well as denotes the different
methods upon which is applied by people from all around the
C. Usage of Biometrics today world ranging from the geometry identification based upon the
Biometric authentication, or just biometrics, is defined person’s hand shape to the usage of techniques such as using
by biometrics.gov [9] as the process of making sure that the K-Nearest Neighbor (KNN) and Multi-Layer perceptron for
person is who he claims to be. Authentication of the identity of Neural[11]
the user can be done in three ways; it’s something that person An article authored by Berghoff et al. explains the different
knows (a password), something the person has (a key or card), possible attacks that can be mounted upon biometrics both with
and something the person is (fingerprints, footprint).Biometrics and without the usage of AI. Elaborating on the details upon
is based on the anatomic uniqueness of a person, and unique
which malicious actors can penetrate various systems, such as
characteristics can be used to prevent unauthorized access by
the usage of a backdoor to falsify the AI model implemented in
using biometric control.There are several classifications of
biometrics, to ‘morphing’ attacks where faces are combined to
biometrics, among them including; DNA, Ear shape, Eye
create a very similar face, taking advantage of a facial
(either the Iris, Retina, or Scleral vein), Face, Finger geometry,
recognition systems ability to recognize variance in a humans
Fingerprint, Gait, Hand Geometry, Heartbeat, Keystrokes
face creating a large difficulty in tackling such attacks. The
(Typing), Odour, Signatures, Vascular (Vein), and Voice
study also suggested the usage of response methods to prevent
Institutions including that of an educational field, have attackers from relying on pre-computed answers, such as the
grown in size substantially over generations which then leads usage of recording video to show the user moving their hand or
to the need for multiple systems of control and regulation. One showing certain objects. A method that recently has been used
of the primary issues for governing bodies and local education by the likes of Lockdown browser before an exam to ensure a
authorities is reconciling a need for bureaucratic efficiency student's facial recognition and attendance [12].
whilst ensuring that the children in their care are where they Virata & Festijo in 2019 [13] explored the usage of
should be, engaged in appropriate activities at the right time, biometrics in the Philippines itself and the perspective of
and being fed at appropriate intervals. Within this space, people’s acceptance centered around a single city. The results
developers have sought to support schools (and monetize
presented that the populace that was sampled was willing to
solutions to any number of management problems) by
accept the usage of biometrics, particularly a face-recognition-
providing any number of digital products to streamline and
based system that would be of large benefit to the country. It is
enhance school processes, for example, for attendance
worth noting in the results that among the variables measured,
monitoring, assessment, and accounting/audit.
there is a strong correlation between the design of the system
A survey conducted by FICO, a software firm specializing with regards to security, indicating a want for trust with the
in analytics presented survey data in 2020 regarding Filipinos system, design with relation to ease of use, showing the appeal
usage of biometrics in banking [10] where 53%, barely more for an intuitive system which appeals to all demographics. This
than half of the survey respondents were using separate implies that the populace is willing to adapt to such a system.
passwords, a remaining 18% reuse passwords, and 14% use the Evaluating the Biometric Attendance System (BAS) at
same password for all accounts, a terrible practice. In turn, Cebu Technological University-Tuburan Campus provides
however, the alternative for passwords and usernames is valuable insights into the practical application and performance
biometrics which 72% were willing to use, most of which of biometric technology in an educational institution during
preferred fingerprint scanning at 76%, 40% for facial 2018 [14]. According to Gabuya et al. (2022), the study,
recognition, and 23% for an eye scan -see figure 1. conducted among teaching and non-teaching staff, emphasizes
key aspects such as timeliness, accuracy, graphical user-
interface, and system maintenance. The findings underscore the
significance of biometric systems in ensuring employee
punctuality, maintaining accurate attendance records, and

44

Authorized licensed use limited to: VTU Consortium. Downloaded on February 24,2025 at 18:28:41 UTC from IEEE Xplore. Restrictions apply.
enhancing user experience through features like a touch-screen Survey Questions Responses
interface and system setup. The identified challenges, such as 1. Are you in the tech Yes- 90%
the common issue of employees forgetting to log in/out, industry? No- 10%
resonate with the real-world complexities of implementing Facial Recognition - 30%
biometric solutions. Furthermore, formulating a proposed 2. Which of these biometrics Fingerprint Scanning - 83.3%
system improvement based on identified lapses reflects the do you use in your daily life? Voice Recognition - 6.7%
dynamic nature of biometric technology and its ongoing Password - 3.3%
relevance in optimizing attendance-tracking systems. This Banking - 33.3%
3. Which part of your life do Personal Devices - 83.3%
research contributes to the broader understanding of how you use biometrics? Workplace - 16.7%
biometrics are utilized in contemporary organizational settings, Online Transcations - 43.3%
shedding light on the strengths and areas for improvement in 4. Do you find biometrics Yes -86.7%
their practical application. convinient over the usual
No- 13.3%
username and password?
Addressing the AI bias in clinical decisions underscores a 5. Are you aware of any Highly Aware -53.3%
critical aspect of the current landscape in biometrics and emerging threats towards
healthcare technology [15]. According to Norori et al. (2021), biometrics such as AI digitally
the issue of algorithmic bias emerges as a significant challenge swapping or reconstructing Somewhat aware- 40%
your face to gain illegal access
as artificial intelligence (AI) plays an increasingly vital role in
or copying your voice?
clinical decision-making processes. This is particularly relevant 6. Do you believe that AI's Yes- 90%
to biometrics, where AI algorithms heavily rely on extensive growth and dev't, the current
datasets. The potential consequences of bias in these datasets safety measures with No- 10%
are highlighted, especially when certain demographic groups biometrics are enough?
are underrepresented or misrepresented. In biometrics, this bias 7. Would you still prefer to use Yes- 50%
biometrics over other security
can lead to serious ramifications, such as misdiagnoses and a No- 43.3%
measures
lack of generalization. The research not only acknowledges this
challenge but also proposes practical steps and tools from the
field of open science to mitigate bias and enhance the fairness Item 1 in table 1 shows that the majority of the participants,
of AI algorithms in clinical decision-making and biometric at 90%, are not in the tech industry or have backgrounds in
applications in healthcare. technology, while the remaining 10% fall under it. Item no. 3
depicts the preference in the type of biometric that they use, 25
Utilizing relational learning on pattern-match rules may not
respondents, or 83.3%, prefer using a fingerprint scanner, 9
have a direct relationship with the usage of biometrics today,
respondents use facial recognition, and 2 prefer voice
however, it focuses on information extraction from natural
recognition. 1 answered none, and another preferred to just use
language documents using machine learning techniques [16].
a password. Moreover, 83.3% of the respondents would use
According to Califf & Mooney (1999), a system, named
biometrics for authenticating on their personal devices, 13 or
RAPIER, is presented. It is designed to induce pattern-match
43.3% for online transactions, 10 or 33.3% for banking, 5 or
rules for extracting relevant information. While the context
16.7% for the workplace, 3 or 10% for healthcare, 1 had
revolves around general information extraction, the
responded with ‘commands’ and the remaining 2 none. Item
methodologies and algorithms developed could potentially find
No .4 illustrates that the majority, 86.7% of respondents, find
application in various domains, including biometrics. In
biometrics convenient over the conventional username and
biometrics, where extracting meaningful patterns and
password, the remaining 13.3% say otherwise. In item 5,
information from diverse datasets is crucial, machine learning
53.3% of respondents demonstrate a moderate level of
approaches for information extraction, like those presented in
awareness of such threats, 40% are highly aware and the
this research, could contribute to enhancing the efficiency and
remaining 6.7% remain minimally aware. Item No. 6 illustrates
accuracy of biometric systems through pattern recognition and
people’s personal take on the question, half say no, while
rule induction techniques.
43.3% say otherwise, the remaining 2 respondents gave
III. METHODOLOGY personal answers. Lastly, item 7 shows half of the respondents
would prefer to use biometrics over other measures, 43.4%
To provide a view into the public’s awareness of the would prefer otherwise, and 2 respondents gave varying
capabilities of AI and their usage of biometrics and provide answers.
assistance to the research, a survey was conducted to around 30
participants, the following provide the results presented in IV. CONCLUSION
graphical form
With the data that has been presented. It can be said that
The demographic profile of the respondents reveals around most respondents are willing to accept biometrics in terms of
86.7% are within the range of 18-30 years old, 2 respondents security despite their varying knowledge of emerging threats
are 31-50, and one each for the below 18 and 50+ range. and measures taken to combat them. Due to being asked
whether the respondent comes from a tech background, it was
Table I. Respondent’s response natural to associate their degree of awareness with their
background, to which the majority had answered they were
‘highly aware’. The majority have taken steps, even if minimal,
to protect themselves, with very few having taken none, half

45

Authorized licensed use limited to: VTU Consortium. Downloaded on February 24,2025 at 18:28:41 UTC from IEEE Xplore. Restrictions apply.
believe that current security methods would maintain [7] McStay, A. (2020). Emotional AI, soft biometrics and the surveillance
protection over personal information and accounts against of emotional life: An unusual consensus on privacy. Big Data & Society,
7(1), 205395172090438 https://doi.org/10.1177/2053951720904386
artificial intelligence-based attacks, and overall, half would still
[8] Michael, K., Abbas, R., Jayashree, P., Bandara, R., & Aloudat, A.
prefer biometrics personally over Multifactor Authentication. (2022). Biometrics and AI bias. IEEE Transactions on Technology and
Advancements in artificial intelligence have made the need Society, 3(1), 2–8. https://doi.org/10.1109/tts.2022.3156405
for better privacy and security higher than before. However, [9] Biometrics | Homeland Security. (2023).
http://www.biometrics.gov/documents/biohistory.pdf
while the current methods are perceived to be useful against
[10] Fintech News Philippines. 2020. FICO Survey: Filipinos showing
spoofing, there is still the looming threat of the likes of greater acceptance for biometrics - Fintech Singapore. Fintech Singapore.
deepfakes and artificial image generation through the scraping https://fintechnews.sg/41682/fintechphilippines/fico-survey-
of data. One proposed solution against such a threat is the filipinosshowing-greater-acceptance-for-biometrics/.
study of ‘cloaking’ the images through software such as [11] Iyer, A.P. et al. (2020) Journal of Critical Reviews an analysis of
Fawkes [22] using changes made that are imperceptible to the artificial Researchgate. Available at:
human eye but are felt by deep learning models making facial https://www.researchgate.net/publication/340443652_Journal_of_Cr
itical_Reviews_AN_ANALYSIS_OF_ARTIFICIAL_INTELLIGENCE
recognition software causing them to misidentify the faces _ IN_BIOMETRICS-THE_NEXT_LEVEL_OF_SECURITY (Accessed:
shown when trained upon. Despite this, however, there is the 16 January 2024).
issue of having such methods spread to the majority of people [12] Berghoff, C., Neu, M., and Von Twickel, A. 2021. The interplay of AI
and are currently limited by a user’s efforts to protect and biometrics: challenges and opportunities. IEEE Computer 54, 9,80–
themselves, made more difficult by the widespread use of 85. https://ieeexplore.ieee.org/document/9524669
social media in the country, creating a large amount of data that [13] Virata, A.J.A. and Festijo, E.D. 2019. Understanding the Acceptance of
can be scraped unless hidden by privacy settings. Due to such a Face Recognition-Based Identity Verification System: An Input to its
Design and Development in the Philippines. 2019 IEEE 4th International
limitations, awareness, and education remain key for the Conference on Computer and Communication Systems (ICCCS).
Philippines to maintain an edge over the increasing usage of https://ieeexplore.ieee.org/document/8821785
artificial intelligence. [14] Gabuya, A. Q., Zosa, L. T., & Minoza, J. T. (2022). The Performance of
Biometric Attendance System (BAS): CTU-Tuburan Campus as case
This paper explores the integration of biometrics, study. International Journal of Scientific and Research Publications,
particularly AI-driven technologies, emphasizing the potential 12(7), 419–426 https://www.researchgate.net/profile/Alden-
benefits and risks associated with this technological Jr-Gabuya/publication/3
advancement, focusing on the security implications, privacy 62506288_The_Performance_of_Biometric_Attendance_System_B
concerns, and the impact of emerging technologies like AS_CTU-Tuburan_Campus_as_case_study/links/62ed4a8c88b83e
7320ae4b91/The-Performance-of-Biometric-Attendance-System-BA S-
deepfakes on biometric data manipulation. The comprehensive CTU-Tuburan-Campus-as-case-study.pdf
resource aims to inform discussions and strategic decision- [15] Norori, N., Hu, Q., Aellen, F. M., Faraci, F. D., & Tzovara, A. (2021).
making in technological security. Addressing bias in big data and AI for health care: A call for open
science. Patterns, 2(10),
REFERENCES 100347https://www.cell.com/patterns/pdf/S2666-3899(21)00202-6.pd
[16] Califf, M. E., & Mooney, R. J. (1999). Relational learning of pattern-
[1] Lai, K., De Oliveira, H. C. R., Hou, M., Yanushkevich, S., & Shmerko, match rules for information extraction. National Conference on Artificial
V. (2020). Risk, Trust, and Bias: Causal Intelligence, 328–334.https://cdn.aaai.org/AAAI/1999/AAAI99-
regulators of Biometric-Enabled Decision Support. IEEE 048.pdf
Access, 8,148779–148792.
https://doi.org/10.1109/access.2020.3015855
[2] Utegen, D., & Rakhmetov, B. (2023). Facial recognition technology and
ensuring security of biometric data: Comparative analysis of legal
regulation models. Journal of Digital Technologies and Law, 1(3), 825–
844. https://doi.org/10.21202/jdtl.2023.36
[3] Hayek, M. (2022). Biometrics and Personal Data as a Security Concern
in the 21st Century? CU Digital Repository. Retrieved January 29,2024,
from
https://dspace.cuni.cz/bitstream/handle/20.500.11956/174442/12041808
1.pdf
[4] Nguyen, H. (2023, August). A Qualitative Exploratory Research Design
Study of Asian American Consumer Acceptance of Biometric
Technology. ProQuest. Retrieved January 29, 2024, from
https://www.proquest.com/openview/ddf10bfe0977064f2608d7813f401
25c/1
[5] Blauth, T. F., Gstrein, O. J., & Zwitter, A. (2022). Artificial Intelligence
Crime:An overview of Malicious use and abuse of AI. IEEE Access, 10,
77110–77122. https://doi.org/10.1109/access.2022.3191790
[6] Bodepudi, A. and Reddy, M. 2020. Spoofing Attacks and Mitigation
Strategies in Biometrics-as-a-Service Systems. Eigenpub Review of
Science and Technology. 4, 1
(Feb. 2020), 1–
14.https://studies.eigenpub.com/index.php/erst/article/view/10/9

46

Authorized licensed use limited to: VTU Consortium. Downloaded on February 24,2025 at 18:28:41 UTC from IEEE Xplore. Restrictions apply.