MODERN APPROACHES TO ENSURING NETWORK SECURITY USING DPI TECHNOLOGY

Abstract

Deep Packet Inspection (DPI) remains an important tool in ensuring network security. Unlike traditional packet filtering, DPI deeply examines the header and payload of packets, which allows detecting complex threats. Today, the increase in network traffic and the proliferation of encrypted connections have imposed new limitations on DPI methods: delays in networks, privacy issues, and the emergence of new types of attacks are observed. In this article, we analyze traditional and advanced approaches to DPI technology, and consider the possibilities of combining them with machine learning (ML). Based on a literature review and comparative analysis, a new methodology is developed and simulation results are presented. Our study shows that a DPI system enriched with artificial intelligence can significantly improve security. In conclusion, we provide recommendations on the limitations of traditional DPI, its relevance, and future research directions, such as encrypted traffic analysis and understandable ML (XAI) methods.