The number of Internet of Things (IoT) devices has increased every day and along with this growth arises the security concerns. Several techniques have been studied for the prevention, detection and treatment of attacks in conventional networks, such as the work of KDD CUP 99 that proposed a labeled collection, which has been quite exploited in recent decades. A good evaluation of techniques and algorithms of intrusion detection systems is related to the existence of good datasets. However, few works exploit the detection of attacks on Internet of Things and until now no collection of data has been proposed for this problem. Along with new technologies and devices arise new techniques of invasion, and even more elaborated. Therefore, it is necessary to treat the attack detection problem in a special way. In view of this, this work is dedicated to setting up a test environment that represents an Internet of Things network, collecting normal device traffic, simulating attacks, assembling a collection of data and analyzing it. For this, we run invasion tests on emulated devices, resulting in a new collection of data. We validate the new collection by applying machine learning algorithms and comparing with the KDD collection.

No abstract

Espera-se que a quantidade de objetos da Internet das Coisas cresça de forma exponencial, aumentando consequentemente o fluxo de dados trafegados pela rede. Para suportar este aumento, a diversidade de tecnologias e a possível distribuição geográfica destes objetos conectados, é necessário uma infraestrutura apta a gerenciar estas demandas. Este trabalho apresenta os detalhes da implantação e os experimentos de integração do SOFTWAY4IoT realizado junto à infraestrutura computacional (máquinas virtuais) disponibilizada pela RNP e a plataforma FIWARE-Lab@RNP. O caso de uso é lixeiras inteligentes para coleta de dados acerca da quantidade de resíduos orgânicos e monitoramento de níveis de lixo das lixeiras seletivas.

No abstract