Detection of Layer 2 Attacks Based on t-SNE Algorithm

Abstract

Artificial intelligence-based methods that focus on detecting attacks on Layer-2 are becoming increasingly important to increase network security. These methods aim to minimize the effects of attacks by identifying potential threats at Layer-2 at an early stage. In particular, optimization-based machine learning algorithms stand out as a powerful tool for detecting complex attacks on Layer-2. Within the scope of the study, data obtained from circuit elements were collected in order to ensure Layer-2 security and feature extraction was performed using the t-SNE algorithm on this data. The t-SNE algorithm reduced high-dimensional data to a lower-dimensional space, allowing better analysis of the relationships and structures between data points. As a result of the analyzes, the features obtained in the feature extraction performed with the t-SNE algorithm were classified with Decision Tree (DT) and a success rate of 90.8% was achieved. This high success rate shows how effective t-SNE is in detecting attacks on Layer-2 and reveals that it can play a critical role in network security. It is envisaged that such AI-based methods may be used more widely in the future to detect and prevent attacks at Layer-2 and other network layers. © 2024 IEEE.