Federated Learning Based Anomaly Network Intrusion Detection System Using RQA

Modern cybersecurity environments face rapidly evolving threats that bypass conventional perimeter-based defenses and signature-only detection mechanisms. This paper presents a modular, multilayered Cyber Intrusion Detection System (Cyber IDS) designed to identify malicious activity through real-time packet sniffing, signature-based inspection, and anomaly detection techniques. By integrating a lightweight sniffer engine, recursive queue analysis (RQA), and dynamic rule-based signature matching, the system provides early detection of suspicious patterns, distributed attacks, and unauthorized access attempts. The methodology includes packet capture, feature extraction, behavioral scoring, and multi-stage detection logic supported by fast database lookup. Through empirical evaluation across diverse traffic profiles, the system demonstrates its capability to detect anomalies with improved precision and lower false positive rates compared to static signature-only IDS models. The proposed solution establishes a scalable foundation for continuous monitoring, adaptive threat detection, and rapid response in modern networked environments.