Behavior of the DenStream Clustering Algorithm for Attack Detection in the Internet of Things

Multiple attack detection schemes based on supervised batch learning are presented in the literature as an alternative to improve Internet of Things (IoT) security. These schemes require benign and malicious traffic samples for training and are unable to easily adapt to changes in the analyzed data....

Descripción completa

Detalles Bibliográficos
Autores: Tazima, Gabriel Keith, Zarpelao, Bruno
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2023
País:Brasil
Institución:Universidade Estadual de Londrina (UEL)
Repositorio:Revista Semina: Ciências Exatas e Tecnológicas (Online)
Idioma:inglés
OAI Identifier:oai:ojs2.ojs.uel.br:article/48956
Acceso en línea:https://ojs.uel.br/revistas/uel/index.php/semexatas/article/view/48956
Access Level:acceso abierto
Palabra clave:stream mining
cyberattack detection
internet of things
cybersecurity
mineração de fluxos contínuos de dados
detecção de ciberataques
internet das coisas
cibersegurança
Descripción
Sumario:Multiple attack detection schemes based on supervised batch learning are presented in the literature as an alternative to improve Internet of Things (IoT) security. These schemes require benign and malicious traffic samples for training and are unable to easily adapt to changes in the analyzed data. In this work, we study how we can use DenStream, an unsupervised stream mining algorithm, to detect attacks in IoT networks. This type of algorithm does not require labeled examples and can learn incrementally, adapting to changes. We aim to investigate whether attacks can be detected by monitoring the behavior of DenStream's clusters. The results showed that DenStream could provide indicators of attack occurrence in TCP, UDP, and ICMP traffic.