Enhanced Cyber Security in IoT Using Deep Belief Network

It is incredible how the entire world is dependent on the data and its creation through every possible medium. With the frequent and regular updates and upgrades in technology, human beings have witnessed the growth of smart devices unlike any and have also seen a significant increase in the application of the Internet of Things, it has been an integral part of daily life. Hence, it would not be surprising when realized that these devices are vulnerable to various malicious attacks. These Internet of Things devices are elemental in both design and security and this design frailty is exploited during firmware attacks and their low-level network security is easily penetrated during the cyber-attacks. One such common cyber-attack is Distributed Denial of Services, and the paper aims to address this against a network of Internet of Things devices. In this study, the authors have suggested a deep learning way of classifying the malign data packets sent by the Internet of Things network. Previous work on the above issue is addressed by deep learning models but the major drawback with them is using a dataset that does not contain data generated by the Internet of Things network. Thus, here make use of the UNSW- 2018 IoT Botnet dataset which contains the samples of real- world Internet of Things network transactions along with the synthetic data. Authors employed a deep learning concept called Supervised Deep Belief Network to meet their aim.