Deep Learning Model For IDS In the Internet of Things

Abstract

Emerging technology makes one's life more comfortable; however, in the Internet of Things, there are a lot of weaknesses like infrastructure, connectivity, network, etc, due to the presence of millions of networked devices that make it difficult to implement safety on each device. Security threats are one of the most important issues recently gaining popularity in IoT, attacks that can cause major disruptions and loss of information within the IoT network. Intrusion Detection System (IDS) has a substantial role in protecting and securing an IoT network through detecting and preventing malicious activities. To develop IDS for timely detection and categorization of cyber threats at the network level, classical machine learning techniques are commonly utilized. However, because malicious attacks are continuously evolving and occurring at extremely large sizes, various problems arise, necessitating a scalable solution. In this paper, a convolutional neural network (CNN) approach, which is a kind of deep learning model for IDS discovery, is developed that is flexible and efficient for detecting and classifying cyber-attacks in IoT networks. The well-applied CNN model on the UNSW-NB15 dataset obtained 100% precision results.