Transguard-rt: A Real-time Transformer-driven Intrusion Detection System Using Live Network Traffic Analytics

Abstract

The Rapid Expansion Of Internet-based Communication And Interconnected Digital Infrastructures Has Significantly Increased Exposure To Advanced Cyber Threats, Including Malware Infiltration, Phishing Attempts, Denial-of-service Attacks, And Unauthorized Access. Traditional Security Mechanisms Such As Signature-based Intrusion Detection And Rule-driven Firewalls Are Often Insufficient To Identify Evolving And Previously Unseen Attack Patterns, Particularly In High-speed And Complex Network Environments. To Address These Limitations, This Research Presents A Real-time Intrusion Detection System That Leverages Live Network Traffic Captured Through NCAP- Based Monitoring And Applies A Transformer-based Deep Learning Architecture For Intelligent Threat Detection. The Proposed Approach Focuses On Analyzing Sequential Dependencies And Contextual Relationships With In Network Traffic Data Using Self-attention Mechanisms, Enabling Effective Identification Of Both Known And Unknown Intrusion Patterns. The System Pre-processes Captured Traffic, Extracts Relevant Features, And Performs Real-time Classification Of Network Behavior In To Normalormalicious Categories. Continuous Monitoring Ensures Immediate Detection And Response To Potential Cyber Threats, Thereby Enhancing Overall Network Resilience. Performance Evaluation Is Conducted Using Standard Metrics Such As Accuracy, Precision, Recall, And F1-score, Demonstrating Strong Detection Capability And Reliability. The Outcomes Indicate That The Proposed Research Offers A Scalable, Adaptive, And Efficient Cyber Security Framework Suitable For Deployment In Critical Domains Such As Finance, Healthcare, Cloud Environments, And Enterprise Networks, Where Real-time Protection And Intelligent Threat Analysis Are Essential.

Keywords