Desenvolvimento de um sistema de classificação de tráfego de rede em planos de dados programáveis com base na linguagem P4

Abstract

This work proposes the development of a system that allows the classification of network traffic in programmable data plans. The system aims to categorize the flows of applications that constitute the traffic, based on the analysis of their characteristics. To this end, the language for programming packet routing plans called P4 was explored, together with the unsupervised machine learning technique called Threshold-Based Clustering. The results showed the validity of the developed system, since they demonstrate the satisfactory accuracy of the generated clusters, while the average processing time of the packets remained at acceptable levels.