Novel Datasets and Traffic Generation Tools for Intrusion Detection in IoT Communications

Datasets constitute the foundation for the training of a robust Intrusion Detection System (IDS). In particular, IDSs could be useful in IoT scenarios, in which resource-constrained devices can not run traditional security software and require a monitoring entity within their network. In this research, we move along two main directions: on one hand, the analysis and description of significant datasets regarding the security of IoT communications; on the other, the exploration of tools that enable their generation. Following this division, we first depict IoT datasets available in the literature and compare them with respect to their most important characteristics according to the literature. We also compare them according to the most popular features that are typically collected in such scenarios, including network and transport layer information, temporal parameters, payload characteristics, and statistical indicators. We then move our attention to the software tools and frameworks which enable the generation of benign or malicious network traffic in various experimental environments and are used in the construction of datasets. We believe this work could serve as a useful starting point to develop novel approaches and techniques in the field.