View Article

  • Endoguard: An Embedded Security Framework For Real-Time Threat Detection And Prevention In Web Applications

  • Department of CSE (Cyber Security), Sri Venkateswaraa College of Technology, Sriperumbudur, Tamil Nadu, India

Abstract

The rapid growth of web applications and Software-as-a-Service (SaaS) platforms has increased the need for strong and intelligent cybersecurity solutions. Traditional security systems such as firewalls, intrusion detection systems, and web application firewalls mainly protect network boundaries, but they often fail to detect modern attacks that exploit application logic, user behavior, and compromised identities. Threats such as account takeovers, credential stuffing, insider misuse, and business logic abuse frequently appear as legitimate actions and bypass conventional defenses.To solve this problem, this paper proposes EndoGuard, an open-source embedded security framework designed for real-time threat detection and prevention inside web applications. Unlike perimeter-based security systems, EndoGuard integrates directly into the application layer using lightweight SDKs and continuously monitors user actions and internal system events. It captures contextual information such as user identity, IP address, device details, and action type to identify suspicious patterns and abnormal behavior.The framework uses PHP, PostgreSQL, and the Fat-Free Framework for lightweight and scalable implementation. A rule-based engine evaluates events against predefined and customizable security rules and assigns a dynamic risk score to classify activities as low, medium, or high risk. Based on the risk level, the system can trigger actions such as account suspension, multi-factor authentication, or administrator review.EndoGuard improves application-level protection by shifting cybersecurity from external defense to internal monitoring and behavioral analysis. The system is developer-friendly, efficient, and suitable for modern web applications, fintech systems, and enterprise platforms.

Keywords

EndoGuard, Embedded Security, Application Security, Cybersecurity, Risk Scoring, Behavioral Analysis, Real-Time Monitoring, Threat Detection, SaaS Security, Web Application Security.

Introduction

× Popup Image

In today’s digital world, web applications and SaaS platforms are used in banking, e-commerce, healthcare, education, and enterprise systems. These applications process large amounts of sensitive data and critical transactions, making them major targets for cyberattacks.Traditional security solutions such as firewalls, IDS, and WAFs mainly focus on protecting the network perimeter. They are effective against known external attacks like SQL injection and cross-site scripting, but they are weak against identity-based attacks and business logic abuse.Modern attackers often use stolen credentials obtained through phishing or credential stuffing. Since these credentials are valid, traditional systems treat these actions as normal. This creates a serious security gap where attackers can perform fraud, unauthorized transactions, and data theft without detection.To overcome these limitations, embedded security is required. Embedded security integrates protection directly inside the application and continuously monitors user behavior. It odellin user actions, devices, IP addresses, and historical activity to detect suspicious behavior in real time.EndoGuard is designed based on this concept. It provides application-level security by monitoring internal activities, assigning risk scores, and responding immediately to threats before major damage occurs.

RELATED WORK

Many studies show the limitations of traditional perimeter-based security systems. OWASP Top 10 highlights broken access control, authentication failures, and insecure design as major risks in modern web applications.Behavioral analytics research explains that monitoring login patterns, device usage, and unusual transactions helps detect advanced threats such as account takeovers and insider attacks.Threat odelling and attack tree approaches show that attackers often exploit business logic rather than technical vulnerabilities. These attacks are difficult to detect using traditional firewalls.Machine learning and anomaly detection systems improve threat detection, but many are complex and expensive to deploy. EndoGuard focuses on a lightweight and practical solution using rule-based detection with future support for intelligent models.

PROPOSED SYSTEM

  1. Existing System

Existing systems mainly rely on authentication, firewalls, and static rule-based detection. Once users are authenticated, their actions are usually trusted without continuous monitoring. This makes systems vulnerable to account hijacking, insider misuse, and logic abuse.

  1. Proposed EndoGuard Framework

EndoGuard is an embedded security framework that works inside the application layer. It monitors user activities such as login attempts, password changes, account updates, and sensitive transactions.

The system follows four major stages:

  1. Event Ingestion
  2. Context Enrichment
  3. Rule Execution
  4. Risk Assessment

Captured events are enriched with contextual data such as IP intelligence, device fingerprinting, and user history. A rule engine then evaluates the event and assigns a dynamic risk score.

  1. Risk Classification
  • Low Risk → Event is logged
  • Medium Risk → Sent for admin review
  • High Risk → Immediate action like MFA or account suspension

This helps reduce false positives and improves response speed.

METHODOLOGY

The system is developed using PHP for backend operations, PostgreSQL for database management, and Fat-Free Framework for lightweight application handling.

Working Process

Figure 1: Workflow Diagram

User Activity

Event Captured by SDK

Context Enrichment

Rule Evaluation

Risk Score Calculation

Low / Medium / High Risk Classification

Security Action Triggered

This workflow explains how every user event moves through the detection pipeline.

  1. User performs an action inside the application
  2. SDK captures the event
  3. Event data is enriched with context
  4. Rule engine processes the event
  5. Risk score is calculated
  6. Security response is triggered
  7. Event is displayed in admin dashboard

The admin dashboard provides audit trails, user timelines, alert queues, and review mechanisms for better monitoring and investigation.

RESULTS AND DISCUSSION

EndoGuard showed strong performance in detecting suspicious activities and reducing delayed threat response.

Key Outcomes

  • Faster identification of abnormal login behavior
  • Better detection of credential misuse
  • Improved monitoring of insider threats
  • Real-time alert generation for high-risk activities
  • Better transparency through audit trails

The system successfully handled event monitoring with low overhead and improved threat visibility compared to traditional external security systems. The risk scoring mechanism helped administrators focus on important threats instead of reviewing every activity manually.

CONCLUSION

EndoGuard provides a practical solution for modern cybersecurity challenges by embedding security directly into web applications. Traditional perimeter-based security is not enough to handle identity-based attacks, insider misuse, and business logic abuse.By combining real-time monitoring, contextual analysis, behavioral tracking, and dynamic risk scoring, EndoGuard improves application-level security significantly. The system is lightweight, scalable, and easy to integrate into existing platforms.This project proves that embedded security frameworks can offer stronger protection than traditional reactive systems and help organizations build safer and more reliable digital platforms.

FUTURE SCOPE

Future improvements for EndoGuard include:

  • Integration of machine learning for advanced anomaly detection
  • Cloud-native deployment support
  • Distributed security architecture for large-scale systems
  • Mobile application monitoring
  • AI-based predictive threat analysis
  • Advanced fraud detection for fintech platforms

These improvements can make EndoGuard more powerful for enterprise-level security environments.

SYSTEM ARCHITECTURE AND DESIGN

  1. Architecture Overview

Figure 1: System Architecture Diagram

The architecture of EndoGuard is designed to provide embedded security directly inside the application layer rather than depending only on external protection systems. The framework is modular, lightweight, and scalable, making it suitable for web applications, SaaS platforms, fintech systems, and enterprise applications.

The architecture mainly consists of the following components:

  1. User Interaction Layer
  2. Event Collection Layer
  3. Context Enrichment Layer
  4. Rule Engine Layer
  5. Risk Scoring Engine
  6. Response and Action Layer
  7. Monitoring Dashboard
  8. Database Management Layer

When a user performs any activity such as login, password reset, fund transfer, profile update, or privilege change, the SDK captures the event and forwards it to the security engine.The context enrichment layer collects additional information such as IP address, browser details, device fingerprint, geolocation, and previous user behavior history. This helps improve decision-making accuracy.The rule engine evaluates the event using predefined and customizable rules. Based on the result, the risk scoring engine calculates a dynamic risk value and classifies the activity.Finally, the response layer performs actions such as logging, administrator alerting, account suspension, or multi-factor authentication.

Figure 2: EndoGuard System Architecture

  1. Module Description

Module 1: User Authentication Monitoring

This module tracks login attempts, failed logins, password changes, and session activities. It helps detect brute-force attacks, credential stuffing, and account takeovers.

Module 2: Context Enrichment Module

This module gathers supporting information such as IP intelligence, device details, browser fingerprinting, and location tracking. It improves anomaly detection by providing context.

Module 3: Rule-Based Detection Engine

This module applies predefined security rules such as unusual login time, repeated failed attempts, sudden device change, and suspicious transaction behavior.

Module 4: Dynamic Risk Scoring Module

Each event is assigned a numerical score based on severity and abnormality. Higher scores indicate higher security risk.

Module 5: Security Response Module

This module triggers actions based on the risk level. Low-risk activities are logged, medium-risk events are flagged for review, and high-risk activities trigger immediate prevention.

Module 6: Admin Dashboard Module

Administrators can monitor events, review alerts, access audit trails, and investigate suspicious users using a centralized interface.

IMPLEMENTATION DETAILS

  1. Software Requirements
  • Operating System: Windows / Linux
  • Frontend: HTML, CSS, JavaScript
  • Backend: PHP
  • Framework: Fat-Free Framework
  • Database: PostgreSQL
  • Server: Apache / XAMPP
  • Browser: Chrome / Edge
  1. Hardware Requirements
  • Processor: Intel i3 or above
  • RAM: Minimum 4 GB
  • Storage: 256 GB or above
  • Internet Connection: Required for deployment and testing
  1. Database Design

Figure 3: Database Design (ER Diagram)

Users

Login Events

Risk Score Table

Alert Management

Admin Review Queue

Audit Trail

This ER-style diagram represents the relationship between core database tables used in EndoGuard.The database stores user information, event logs, risk scores, device history, alerts, and administrator reviews.

Important tables include:

  • Users Table
  • Login Events Table
  • Device Fingerprint Table
  • Risk Score Table
  • Alert Management Table
  • Admin Review Queue

Each event is stored with timestamp, IP address, device details, and action type for future analysis and auditing.

  1. Security Implementation

The system implements:

  • Secure session management
  • Multi-factor authentication triggers
  • Device fingerprint validation
  • IP reputation checking
  • Login anomaly detection
  • Sensitive transaction monitoring
  • Detailed audit trail maintenance

These mechanisms ensure both preventive and detective security.

Figure 2: EndoGuard Deployment Diagram

PERFORMANCE ANALYSIS

Figure 4: Risk Classification Chart

Low Risk → Log EventMedium Risk → Admin ReviewHigh Risk → MFA / Account Suspension

This diagram explains how the system responds based on calculated risk levels.

Figure 5: Admin Dashboard Monitoring Flow

Alerts Generated → Review Queue → Admin Analysis → Decision → Final Action

This shows how administrators handle suspicious events inside the monitoring dashboard.

  1. Event Processing Performance

The system processes user events in real time with minimal latency. Event capture and risk evaluation happen instantly after user interaction, ensuring immediate threat visibility.

  1. Threat Detection Accuracy

Testing showed improved detection of:

  • Account takeover attempts
  • Credential stuffing attacks
  • Unusual login behavior
  • Insider misuse
  • Business logic abuse

Compared to traditional systems, EndoGuard provided better accuracy because it considered user behavior and context instead of only network traffic.

  1. Risk Scoring Efficiency

The dynamic scoring system reduced unnecessary alerts and helped administrators prioritize serious incidents first. This improved operational efficiency and reduced false positives.

  1. Response Time

High-risk activities triggered immediate responses such as MFA requests and account restrictions. This proactive model reduced the damage window significantly.

  1. Reliability and Scalability

The lightweight architecture allowed stable performance even with increasing event volume. The modular structure supports future cloud deployment and enterprise-scale expansion.

ADVANTAGES OF PROPOSED SYSTEM

  1. Real-time monitoring of user activities
  2. Strong protection against account takeover attacks
  3. Detection of insider threats and privilege misuse
  4. Reduced false positives using contextual analysis
  5. Faster response to suspicious activities
  6. Lightweight and easy integration
  7. Open-source and cost-effective implementation
  8. Improved transparency through audit logs
  9. Better administrator control through dashboard monitoring
  10. Scalable for modern SaaS and enterprise platforms

LIMITATIONS

Although EndoGuard provides strong application-level protection, some limitations still exist:

  1. Current implementation mainly supports small and medium-scale systems
  2. Rule-based detection depends on proper rule design
  3. Full machine learning integration is not yet implemented
  4. Large-scale cloud-native deployment requires future enhancement
  5. Advanced fraud prediction requires larger historical datasets

These limitations can be addressed in future versions.

REFERENCES

  1. OWASP Foundation, “OWASP Top 10: The Ten Most Critical Web Application Security Risks,” 2021.
  2. Ross Anderson, “Security Engineering: A Guide to Building Dependable Distributed Systems,” 2020.
  3. Bruce Schneier, “Attack Trees: Modeling Security Threats,” 2015.
  4. Adam Shostack, “Threat Modeling: Designing for Security,” 2014.
  5. Gary McGraw, “Software Security: Building Security In,” 2006.
  6. Matt Bishop, “Computer Security: Art and Science,” 2018.
  7. A. B. Johnston and A. Gupta, “The Role of Behavioral Analytics in Cybersecurity,” 2019.
  8. E. Alpaydin, “Introduction to Machine Learning for Cybersecurity Applications,” 2020.
  9. S. Garfinkel and G. Spafford, “Web Security, Privacy and Commerce,” 2016.
  10. P. Samarati and S. de Vimercati, “Access Control: Policies, Models, and Mechanisms,” 2001.

Reference

  1. OWASP Foundation, “OWASP Top 10: The Ten Most Critical Web Application Security Risks,” 2021.
  2. Ross Anderson, “Security Engineering: A Guide to Building Dependable Distributed Systems,” 2020.
  3. Bruce Schneier, “Attack Trees: Modeling Security Threats,” 2015.
  4. Adam Shostack, “Threat Modeling: Designing for Security,” 2014.
  5. Gary McGraw, “Software Security: Building Security In,” 2006.
  6. Matt Bishop, “Computer Security: Art and Science,” 2018.
  7. A. B. Johnston and A. Gupta, “The Role of Behavioral Analytics in Cybersecurity,” 2019.
  8. E. Alpaydin, “Introduction to Machine Learning for Cybersecurity Applications,” 2020.
  9. S. Garfinkel and G. Spafford, “Web Security, Privacy and Commerce,” 2016.
  10. P. Samarati and S. de Vimercati, “Access Control: Policies, Models, and Mechanisms,” 2001.

Photo
Dhanuja V.
Corresponding author

Department of CSE (Cyber Security), Sri Venkateswaraa College of Technology, Sriperumbudur, Tamil Nadu, India

Photo
Gopika U.
Co-author

Department of CSE (Cyber Security), Sri Venkateswaraa College of Technology, Sriperumbudur, Tamil Nadu, India

Photo
Kuraku Vamsidhar
Co-author

Department of CSE (Cyber Security), Sri Venkateswaraa College of Technology, Sriperumbudur, Tamil Nadu, India

Dhanuja V.*, Gopika U., Kuraku Vamsidhar, Endoguard: An Embedded Security Framework For Real-Time Threat Detection And Prevention In Web Applications, Int. J. Sci. R. Tech., 2026, 3 (5), 264-270. https://doi.org/10.5281/zenodo.20080119

More related articles
PSO_KAN: A Hybrid Particle Swarm Optimization and ...
Umar Kabir Umar, Maryam Alka, Ogochukwu John Okonko, Rumana Kabir...
Vigilance-V: An AI-Powered Real-Time Access and Be...
Mayur Gavali, Adityaraj Gaikwad, Vedika Koravi, Shreya Kamble, Af...
A Review on Network Intrusion Detection...
Hiba Fathima K P, Anugraha P P...
Antibiotic Resistance – A Threat to Public Health – A Review...
Muhil K. S., Yoka T., Selvakumar, Mahashree B., Keerthivasini C., Madesh P....
Privacy and Cybersecurity in Smart Devices: Challenges and Opportunity...
Oketayo Abimbola M., Oduwole Oluwakemi O., Nriagu’ Chukwunonso...
Related Articles
Real-Time Xai Framework For Early Behavioural Ransomware Detection With Threat I...
Bachina Parimala, Srilakshmi Dupati, Hema Karpurapu, Latha Gaddam...
Cyber Security Awareness Among Genz...
J. Anurushmitha, J. P. Pramod, D. Nikhitha...
PSO_KAN: A Hybrid Particle Swarm Optimization and Kolmogorov Arnold Network for ...
Umar Kabir Umar, Maryam Alka, Ogochukwu John Okonko, Rumana Kabir Aminu, Idris Yau Idris, Fatima Shi...
More related articles
PSO_KAN: A Hybrid Particle Swarm Optimization and Kolmogorov Arnold Network for ...
Umar Kabir Umar, Maryam Alka, Ogochukwu John Okonko, Rumana Kabir Aminu, Idris Yau Idris, Fatima Shi...
Vigilance-V: An AI-Powered Real-Time Access and Behavioral Analytics Platform...
Mayur Gavali, Adityaraj Gaikwad, Vedika Koravi, Shreya Kamble, Affan Kotwal...
A Review on Network Intrusion Detection...
Hiba Fathima K P, Anugraha P P...
PSO_KAN: A Hybrid Particle Swarm Optimization and Kolmogorov Arnold Network for ...
Umar Kabir Umar, Maryam Alka, Ogochukwu John Okonko, Rumana Kabir Aminu, Idris Yau Idris, Fatima Shi...
Vigilance-V: An AI-Powered Real-Time Access and Behavioral Analytics Platform...
Mayur Gavali, Adityaraj Gaikwad, Vedika Koravi, Shreya Kamble, Affan Kotwal...
A Review on Network Intrusion Detection...
Hiba Fathima K P, Anugraha P P...