Análise de desempenho com algoritmos de grupos de frequência aplicados em sistemas autonômicos para proteção de redes de computadores
On the context of autonomic computing, there is a system called Of-IDPS that aims to detect and react to attacks from a network, by analyzing your usage history and security alerts autonomously, with the least possible human intervention. For this, in its architecture, the Of-IDPS depends of a unsup...
| Autor principal: | Morais, Vinícius Ribeiro |
|---|---|
| Formato: | Trabalho de Conclusão de Curso (Graduação) |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2020
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/6034 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
On the context of autonomic computing, there is a system called Of-IDPS that aims to detect and react to attacks from a network, by analyzing your usage history and security alerts autonomously, with the least possible human intervention. For this, in its architecture, the Of-IDPS depends of a unsupervised learning algorithm, related to frequent items mining, to generate security rules that are able to mitigate attacks that may affect the network through the generated security rules. Therefore, our work objective aims to improve Of-IDPS performance with the usage of differents items frequent mining algorithms, trying to improve the response time of Of-IDPS and consequently helping in the action against cyber threats. To analyze the performance, the evaluation of this algorithms was made using metrics like time and amount of memory spent in the execution of the algorithms. To verify the purpose of this research, the algorithms were submitted to synthetic databases, to be evaluated and preselected. After the preselection, the algorithms that got the best results were applied in the Of-IDPS, to be analyzed in a network scenario. In the experiments, the results obtained indicated the best algorithms, being these: Apriori, LCMFreq and FP-Growth. Applying these algorithms in the Of-IDPS resulted a 26% improvement in containment of malicious packets with the LCMFreq in comparison to FP-Growth. Besides that, the LCMFreq mitigated 81.81% of malicious packets in an analysis of the network scenario with and without the Of-IDPS, in other words, with the LCMFreq the mitigation of malicious packets was bigger and more faster than other algorithms. Thus, it was possible to affirm that there were improvements in the performance of the IDPS with the use of new algorithms of frequent items. |
|---|