Emerging Trends of Recently Published Datasets for Intrusion Detection Systems (IDS): A Survey

TL;DR

This survey reviews recent datasets used for training and evaluating Intrusion Detection Systems, highlighting their importance in evolving cybersecurity threats and emphasizing the need for more datasets for emerging technologies.

Contribution

It provides a comprehensive review of seven recent IDS datasets, develops a taxonomy for dataset selection, and discusses future needs for datasets in IoT and mobile security.

Findings

Seven recent datasets reviewed for IDS research

A taxonomy for dataset selection is proposed

Highlights the need for datasets for IoT and smartphones

Abstract

With the ubiquitous nature of information technology solutions that facilitate communication in the modern world, cyber attacks are increasing in volume and becoming more sophisticated in nature. From classic network-based Denial of Service (DoS) attacks to the more recent concerns of privacy compromises, Intrusion Detection Systems (IDS) are becoming an urgent need to safeguard the modern information technology landscape. The development of these IDS relies on training and evaluation datasets that must evolve with time and represent the contemporary threat landscape. The purpose of this analysis is to explore such recent datasets, describe how they enable research endeavours and the development of novel IDS. Specifically, 7 recent datasets published for IDS research have been reviewed along with selected publications that have employed them. In doing so, the discussion emphasizes the need for the publication of even more modern datasets, especially for emerging technologies such as the Internet of Things (IoT) and smartphone devices, to ensure that modern networks and communication channels are secured. Furthermore, a taxonomy based on the discussed datasets has been developed that can be used to inform the dataset selection process for future research endeavours.