The evolution of computing devices has allowed the evolution of
service provision in society, applying new technologies based on
the Internet of Things (IoT). Most IoT devices have security vulnerabilities,
making them susceptible to Distributed Denial of Service
(DDoS) Attacks. Thus, it is necessary to apply solutions that can
detect this type of attack in IoT networks from the information of
the network traffic. However, there is still no definition of which
traffic characteristics should be used for detection, since the use
of inappropriate characteristics tend to make detection difficult.
Within this context, this article presents an analysis of the most important
traffic characteristics for detecting DDoS in IoT networks,
in order to support a detection mechanism based on Machine Learning.
Experiments using a real data set suggest that the proposed
mechanism has an accuracy close to 99 % when the most suitable
characteristics are selected.