INTRUSION RESPONSE THROUGH DEEP PACKET INSPECTION USING MULTI-AGENT SYSTEMS

Abstract

As cyber threats continue to increase in both sophistication and frequency, alongside the growing reliance on cloud-based data storage and networked systems, the need for robust and adaptive security solutions has become increasingly critical. Recent advancements in artificial intelligence (AI) have demonstrated significant potential in automating complex tasks and enabling dynamic, context-aware decision-making, making AI a natural fit for modern cybersecurity applications. By integrating AI-driven techniques into cybersecurity systems, organizations can proactively identify and intercept potential attacks, such as network-based intrusions, before they compromise devices, services, or access points. This thesis proposes a multi-agent intrusion response architecture that integrates LLM-assisted reasoning with traditional classification models to detect and respond to malicious network traffic.