A systematic review on Deep Learning approaches for IoT security

The constant spread of smart devices in many aspects of our daily life goes hand in hand with the ever-increasing demand for appropriate mechanisms to ensure they are resistant against various types of threats and attacks in the Internet of Things (IoT) environment. In this context, Deep Learning (DL) is emerging as one of the most successful and suitable techniques to be applied to different IoT security aspects. This work aims at systematically reviewing and analyzing the research landscape about DL approaches applied to different IoT security scenarios. The contributions we reviewed are classified according to different points of view into a coherent and structured taxonomy in order to identify the gap in this pivotal research area. The research focused on articles related to the keywords 'deep learning', 'security' and 'Internet of Things' or 'IoT' in four major databases, namely IEEEXplore, ScienceDirect, SpringerLink, and the ACM Digital Library. We selected and reviewed 69 articles in the end. We have characterized these studies according to three main research questions, namely, the involved security aspects, the used DL network architectures, and the engaged datasets. A final discussion highlights the research gaps still to be investigated as well as the drawbacks and vulnerabilities of the DL approaches in the IoT security scenario. (C) 2021 Elsevier Inc. All rights reserved.