Page 851
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026
Scamshield – Catch Scammers with Autorecorded Calls: Review
Paper
Shreyanshi Srivastava, Sweta Verma, Pooja Yadav
Information Technology Shri Ramswaroop Memorial College of Engineering & Management Lucknow
(AKTU) Lucknow, India
DOI:
https://doi.org/10.51583/IJLTEMAS.2026.150400077
Received: 10 April 2026; Accepted: 15 April 2026; Published: 09 May 2026
ABSTRACT
The rise of digital communication has been accompanied by a significant increase in fraudulent and scam
phone calls, causing both financial and emotional damage to individuals and organizations. Traditional rule-
based detection systems have become insufficient in combating these evolving scams, which often rely on
voice manipulation and social engineering techniques. This review paper explores recent research and
technological advancements in artificial intelligence (AI), voice biometrics, and data mining methods for scam
and fraud call detection. It surveys existing literature in the domain of telecommunication fraud prevention,
highlighting methods such as machine learning–based classification, real-time voice recognition, and
behavioral pattern analysis. The proposed system, ScamShield, integrates these ideas to create a lightweight
Android-based application capable of identifying suspicious calls through voice and keyword analysis, storing
call metadata, and providing scam-awareness alerts. The review aims to bridge existing research gaps by
summarizing multiple approaches to telecommunication fraud detection and offering insights for future AI-
powered solutions that ensure secure and trustworthy communication networks.
Keywords: Telecommunication Fraud, Scam Call Detection, Artificial Intelligence, Voice Recognition,
Machine Learning, Real-Time Detection, ScamShield, Fraud Prevention, Speech Processing, Keyword
Analysis.
INTRODUCTION
The global telecommunication industry has transformed the way individuals communicate and access
information. While this revolution has improved convenience and connectivity, it has also created an avenue
for cybercriminals to exploit users through fraudulent phone calls, impersonation scams, and voice phishing
attacks. According to the Communications Fraud Control Association (CFCA), the global loss from telecom
fraud exceeded USD 38 billion in 2023. This growing trend reflects the inability of traditional call blocking
and number-based reporting mechanisms to deal with new, technology-driven scam patterns.
Existing spam filters primarily depend on user reports or community databases such as Truecaller. However,
scammers frequently use dynamic phone numbers, spoofed caller IDs, and VoIP-based networks, making static
detection models ineffective. Many victims, especially elderly and rural populations, fall prey to emotional
and psychological manipulation during calls, highlighting the need for proactive scam detection and
prevention mechanisms.
The proposed project, ScamShield, addresses this issue by integrating real-time call monitoring, keyword-
based detection, and secure Firebase storage into a single Android application. The app continuously observes
call states, identifies potential scam-related patterns, and stores call details for future analysis or reporting. Unlike
complex enterprise-level systems, ScamShield is designed to be lightweight, user-friendly, and privacy-
conscious, making it ideal for personal mobile devices.
Page 852
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026
The objective of this review paper is to summarize key research contributions in fraud call detection, analyze
modern approaches like behavioral monitoring and speechbased identification, and explain how ScamShield
adapts these ideas into a practical mobile system.
Fraudulent communication has become a sophisticated cyber threat, evolving beyond simple spam calls into
organized networks employing automation, spoofing, and voice simulation. Scammers now use technologies
such as caller ID spoofing and automated speech systems to mimic trusted institutions like banks or government
agencies. These evolving tactics make traditional rule-based filters inadequate for accurate scam identification.
Hence, analytical and content-based monitoring models are now required to process call data, recognize scam-
related speech, and classify it intelligently.
Although machine learning and artificial intelligence have shown great promise in large-scale telecom fraud
detection, such solutions are often cloud-dependent and complex for end users. ScamShield bridges this gap by
adapting the conceptual principles of fraud detection into a mobile prototype that performs local monitoring,
data logging, and secure evidence storage directly on the device. This makes the system simple to implement
while laying the groundwork for future AI or NLP integration.
Moreover, the introduction of cloud-integrated frameworks such as Firebase enhances data reliability and
scalability. ScamShield utilizes Firebase Firestore and Storage to securely log call details, metadata, and potential
scam indicators. This not only allows for future enhancement using smarter models but also provides a digital
audit trail for reporting fraudulent cases to authorities. In doing so, the application serves as both a defensive
and investigative tool — protecting users in real time while contributing valuable data toward broader fraud
analysis research.
Finally, with the exponential rise in smartphone adoption, India presents a critical environment for such
innovations. The country ranks among the top victims of phone-based scams, with millions affected annually
through impersonation and financial frauds. ScamShield aligns with India’s Digital Safety Mission and supports
user awareness through automated fraud alerts, secure data handling, and privacy-first design. This makes the
project not only a technical innovation but also a social contribution toward safer digital communication
ecosystems.
LITERATURE REVIEW
The increasing complexity of telecommunication fraud has driven researchers to explore intelligent, data-driven
approaches for fraud detection and prevention. Earlier systems primarily relied on statistical models and call
pattern monitoring, but recent advancements in artificial intelligence (AI) and machine learning (ML) have
significantly improved detection accuracy. Studies such as those by Saloni Malhotra et al. (2023) and Batoul
Abo Yehya et al. (2023) highlight how AI-based systems analyze call metadata, frequency patterns, and user
behavior to identify anomalies in real time. Similarly, Khalid Hafiz Mir et al. (2023) emphasize that real-time
data mining and anomaly detection techniques can effectively recognize suspicious activities within telecom
networks. These approaches form the conceptual backbone of the ScamShield application, which integrates real-
time monitoring, keyword-based filtering, and cloud-based storage for fraud analysis.
Overview of Telecommunication Fraud
Becker et al. (2010) identified fraudulent communication as a major issue in mobile networks, emphasizing the
importance of automated monitoring for unusual call activity. Ferreira et al. (2006) extended this research by
integrating behavior analysis with detection rules to improve fraud recognition accuracy.
Real-Time Detection Approaches
Real-time call monitoring was discussed by Batoul Abo Yehya and Nazih Salhab (2023), who stressed the
importance of identifying suspicious activity as it occurs. ScamShield employs a similar strategy using Android
broadcast receivers to track live call status updates.
Page 853
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026
Behavioural Pattern Analysis
Abidogun (2005) demonstrated that analysing call duration and frequency can reveal hidden fraud indicators.
This concept is reflected in ScamShield’s call monitoring feature, which stores and reviews metadata for potential
anomalies.
Data Mining for Fraud Detection
Research by Becker et al. (2010) and Sandhya et al. (2020) proved that pattern mining helps detect irregular
activity, even with partial call information. ScamShield applies this
by examining basic call records stored in SQLite
AI in Telecommunication Fraud Prevention
Ritika and Mohana (2022) proposed using AI for automatic fraud management and evidence preservation.
ScamShield uses Firebase integration to securely upload suspicious call details, combining automation with
human awareness.
Cloud and Local Database Synchronization
A hybrid data approach combining cloud storage and local caching was recommended by Ritika et al. (2022).
Similarly, ScamShield integrates Firebase (for online data) and SQLite (for offline logs) to ensure reliability even
during disconnection
Prototype Models for Fraud Detection Applications
Prototype models are early-stage implementations used to test the feasibility of fraud detection systems. Studies
like Ritika H. J. and Mohana (2022) and Sameer Qayyum et al. (2010) developed small-scale prototypes to
monitor call data and detect suspicious activities. In ScamShield, the prototype validates essential functions such
as real-time call monitoring, data storage, and scam reporting before integrating advanced AI techniques.
Voice-Based Fraud Systems
Sonwane et al. (2024) introduced the TrustCaller model, which utilized voice recognition to detect
impersonation. While ScamShield doesn’t record or compare voices, it applies the idea of voice-interaction
monitoring for scam detection.
Social Engineering Awareness
Mir et al. (2024) explored how fraudsters exploit emotion and urgency. ScamShield builds on this by introducing
keyword-based scam detection — for instance, triggering alerts for words like “OTP,” “bank,” or “verification.”
Anomaly-Based Call Monitoring
Yehya and Salhab (2023) proposed anomaly-based detection that focuses on deviations from normal usage
patterns rather than predefined blacklists. ScamShield integrates this principle by monitoring all incoming and
outgoing call events dynamically.
Content-Based Scam Identification
Zhao et al. (2018) suggested understanding the content of voice communication for detecting scams. ScamShield
aligns with this through potential transcription-based keyword matching as a future enhancement.
Page 854
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026
Applications
1. Real-time Scam Detection: Monitors ongoing calls and identifies potential scam patterns instantly.
2. Evidence Recording: Stores suspicious call details and recordings in Firebase for later analysis.
3. User Awareness: Alerts users during calls if keywords or scam indicators are detected.
4. Fraud Reporting: Allows easy reporting of scam calls for security or awareness purposes.
5. Call Log Analysis: Maintains a secure database of past calls for identifying frequent scam sources.
6. Data Synchronization: Ensures secure cloud storage and retrieval of evidence for user safety.
7. Lightweight Android Integration: Works smoothly on low-end devices without high resource usage.
8. Expandable Framework: Can integrate future AI or NLP models for automatic scam classification.
Comparision Table
Limitation
The system currently focuses on scam detection and alerting, not automatic call blocking.
Page 855
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026
SUMMARY TABLE
• OS restrictions (especially on iOS) limit background recording and analysis.
• Speech recognition accuracy may drop for regional accents or noisy environments.
• Internet is required for reporting and database synchronization.
• Keyword-based detection can cause false positives or miss new scam patterns.
• Limited dataset and privacy concerns may affect large-scale deployment.
Future Directions
Future developments of ScamShield aim to enhance its detection intelligence through advanced voice and text
analysis. The integration of voice-to-text transcription will allow the system to identify scam-related terms or
suspicious speech patterns in real time without recording full conversations. Additionally, incorporating
emotion recognition could enable the detection of manipulative tones, urgency, or stress cues often used by
Aspect
Existing
Systems
Limitations
ScamShield Solution
Gap Addressed
Call
Detection
Approach
Relies mainly on
user reports and
static spam lists
Real-time monitoring of
ongoing calls with
keywordbased detection
Enables proactive scam
identification instead of reactive
reporting
Data Source
Uses shared global
databases that may
contain outdated or
irrelevant numbers
Uses live user data and call
activity for accurate
monitoring
Ensures updated and personalized fraud
detection
Evidence
Handling
No call recording or
proof mechanism
Stores metadata and
recordings securely on
Firebase
Provides verifiable evidence for
analysis or reporting
Privacy and
Security
Shares user data
publicly for spam
classification
Maintains private cloud
storage linked
to user account
Protects sensitive information while
enabling detection
Scam
Detection
Accuracy
Limited due to
spoofed numbers and
VoIP masking
Detects suspicious
keywords and voice cues
during live calls
Increases accuracy through behavior and
content-based analysis
Offline
Functionality
Requires continuous
internet access
Can function partially
offline for call monitoring
Improves
usability in lowconnectivity areas
User
Awareness
No in-call
alerts for scam
suspicion
Displays realtime alerts
when scam-
like behavior is detected
Prevents user manipulation during
ongoing calls
AI &
Minimal
Designed for
Enables future
Page 856
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026
scammers. These improvements would help ScamShield move toward a more adaptive, context-aware fraud
detection model capable of learning from evolving scam tactics.
Automation
automation and
manual tagging
future integration of AI
and NLP models
Usability
Focused on general
spam identification
Lightweight, user-
friendly mobile app for
personal protection
Reporting
& Analysis
Limited feedback
loop for law
enforcement
Structured database for
fraud reporting and
trend study
CONCLUSION
ScamShield offers a modern, AI-driven approach to scam call detection by analyzing call content and keywords
in real time. It goes beyond traditional number-based filters to provide proactive protection and user awareness.
Though challenges like OS limits, data privacy, and accuracy persist, ScamShield demonstrates strong potential
for creating a secure, intelligent, and userfriendly solution for telecom fraud prevention.
REFERENCES
1. Khalid Hafiz Mir, Ravin Kumar, and Sangeeta Rai, “Real-Time Anomaly Detection in
Telecommunications: Advanced Data Mining Techniques for Fraud Identification,” IEEE Conference on
Data Science and Communication Systems, 2023.
2. Saloni Malhotra, Ginni Arora, and Ruchika Bathla, “Detection and Analysis of Fraud Phone Calls using
Artificial Intelligence,” IEEE ICACCS, May 2023.
3. Batoul Abo Yehya and Nazih Salhab, “Telecommunications Fraud Machine Learning-based Detection,”
IEEE ICAISC, Oct. 2023.
4. Ritika H. J. and Mohana, “Fraud Detection and Management for Telecommunication Systems using
Artificial Intelligence (AI),” IEEE ICCCIS, Oct. 2022.
5. Sameer Qayyum et al., “Fraudulent Call Detection for Mobile Networks,” IEEE ICIET, June 2010.
6. Osama Mohamed Elrajubi et al., “Detection of Bypass Fraud Based on Speaker Recognition,” IEEE
International Conference on Advanced Computing, May 2017.
7. Rushikesh Sonwane et al., “TrustCaller – Voice-based Fraud Prevention System,” IEEE ICACCS, June
2024.
8. Fawcett, T. and Provost, F., “Adaptive Fraud Detection,” Data Mining and Knowledge Discovery, 1(3),
291–316, 1997.
9. H. Weng et al., “Online E-Commerce Fraud: A LargeScale Detection and Analysis,” IEEE ICDE, 2018.
10. S. M. Gowri et al., “Detection of Telephony Spam and Scams using Recurrent Neural Network (RNN)
Algorithm,” IEEE ICACCS, 2021.
11. Abidogun, O. A., “Data Mining, Fraud Detection and Mobile Telecommunications: Call Pattern Analysis
with Unsupervised Neural Networks,” University of the Western Cape Dissertation, 2005.
12. S. Sandhya, N. Karthikeyan, and R. Sruthi, “Machine Learning Method for Detecting and Analysis of
Fraud Phone Calls Datasets,” IJRTE, 2020.
13. Crawford, M. et al., “Survey of Review Spam Detection using Machine Learning Techniques,” Journal
of Big Data, 2015.
14. A. Marzuoli et al., “Uncovering the Landscape of Fraud and Spam in the Telephony Channel,” IEEE
ICMLA, 2016.
15. Luis Cortesao et al., “Fraud Management Systems in Telecommunications: A Practical Approach,”
CiteseerX Journal, 2018.
Page 857
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue IV, April 2026
16. Qianqian Zhao et al., “Detecting Telecommunication Fraud by Understanding the Contents of a Call,”
Cybersecurity Journal, 2018.
17. Kasra Babaei et al., “A Study of Fraud Types, Challenges, and Detection Approaches in
Telecommunication,” Journal of Information Systems and Telecommunication, 2019.
18. Richard A. Becker et al., “Fraud Detection in Telecommunications: History and Lessons Learned,”
Technometrics, 2010.
19. Y. Moreau et al., “A Hybrid System for Fraud Detection in Mobile Communications,” ESANN
Proceedings, 1999.
20. P. Ferreira et al., “Establishing Fraud Detection Patterns Based on Signatures,” Industrial Conference on
Data Mining, 2006.
21. Alsaify, Baha et al., “Voice-Based Human Identification Using Machine Learning,” IEEE ICICS, 2022.
22. Khan M.K. Amjad and Aithal, S., “Voice Biometric Systems for User Identification and Authentication
– A Literature Review,” IJAEML, 2022.
23. Dhakal, P. et al., “A Near Real-Time Automatic Speaker Recognition Architecture for Voice-Based User
Interface,” Machine Learning Applications Journal, 2019.
24. K. S. G. et al., “Voice Comparison Approaches for Forensic Application: A Review,” IEEE ICSCCC,
2023.
25. Z. K. Abdul and A. K. Al-Talabani, “Mel Frequency Cepstral Coefficient and Its Applications: A
Review,” IEEE Access, 2022.
26. Nishtha Tandel et al., “Voice Recognition and Voice Comparison using Machine Learning Techniques: A
Survey,” IEEE ICACCS, 2020.
27. S. Kinkiri and S. Keates, “Speaker Identification: Variations of a Human Voice,” IEEE ICACCE, 2020.
28. H. Kilinc, “A Case Study on Fraudulent User Behaviors in the Telecommunication Network,” Electrica
Journal, 2021.
29. Y. Alraouji and A. Bramantoro, “International Call Fraud Detection Systems and Techniques,” MEDES
Conference, 2014.
30. M. Sahin and A. Francillon, “Understanding and Detecting International Revenue Share Fraud,” Network
and Distributed System Security Symposium, 2021.