Detecção de DDoS Através da Análise da Recorrência Baseada na Extração de Características Dinâmicas
Marcelo Antonio Righi, Raul Ceretta Nunes · 2015
With the increasing number of Distributed Denial of Service (DDoS) attacks, detect them has become essential to maintaining the reliability of institutions using the internet. In this sense, different algorithms have been used to analyze network traffic, such as neural networks, decision trees, principal component analysis and others. However, these algorithms do not use dynamic features to classify network traffic. This article proposes to use the Analysis Quantification of Recurrence based on the extraction of dynamic characteristics combined with the clustering algorithm A-Kmeans to perform traffic classification. The results confirm the accuracy of the model that reached minimal number of false alarms when tested with the CAIDA data set.