Year: 2026
Type: article
Abstract: With the rapid growth of digital networks, managing network traffic efficiently and securely has become increasinglycomplex. Traditional rule-based methods often fail to adapt to dynamic traffic patterns, especially with the widespread adoptionof encryption technologies. In this context, machine learning (ML) has emerged as a powerful solution, offering advancedcapabilities in traffic classification, anomaly detection, and network optimization.This paper provides a detailed review of ML-based approaches, including supervised learning, unsupervised learning, deeplearning, and graph-based techniques. It examines how these methods contribute to improved network performance andsecurity. Additionally, key challenges such as data imbalance, real-time implementation constraints, and computationalcomplexity are discussed in detail. The study brings together insights from various research works, highlighting the growingimportance of AI-driven models in enhancing cybersecurity, predicting traffic behavior, and maintaining quality of service(QoS). It also identifies future research opportunities, including the development of hybrid learning models, the adoption offederated learning, and the integration of ML techniques with modern networking frameworks like Software-DefinedNetworking (SDN) and 5G technologies.
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Keywords: network traffic classification, machine learning for network security, anomaly detection, encrypted traffic, supervised learning, unsupervised learning, deep learning, class imbalance, real-time implementation, computational complexity, quality of service, federated learning, software-defined networking, 5G, hybrid learning model
Open Access status: hybrid
OpenAlex ID: W7217449406
