A Comprehensive Review on Intrusion Detection Systems for Internet of Things

Our way of life makes extensive use of the Internet of Things (IoT). The gadgets connected to the internet, sensors and actuators like smartphones, smart watches, and household goods like smart televisions, air conditioners, refrigerators, sensor lights, sensor curtains, industrial tools connected with sensors and actuators, the smart grid, medical implantable devices, devices used to build smart cities like sensors street lights, garbage monitoring tools, smart vehicles, environment monitoring and others play an important role. In recent years, the IoT has been the focus of significant scientific interest. One component of the IoT is made up of billions of smart, sensor devices that can transfer information over the network. Concern with the protection and privacy of the information always draws the eye. It’s necessary to safeguard the associated network devices and data generated through it from intruders and their malicious programs. Businesses are spending more money on research to improve the way in which these attacks are detected. In an effort to secure IoT devices and their associated data, various rigorous research studies have been conducted to counteract these issues, evade the risks, or minimize the impact on IoT devices and associated data. This chapter is organized into four sections. The first segment illustrates IoT definitions and applications of IoT. In the second section we review the literature work done by researchers to secure the IoT network from various intrusion attacks. In the third segment we briefly discuss the key limitations of IoT devices. The fourth section examines the safety concerns of IoT network from multiple angles. This contains a brief discussion about the Intrusion Detection System (IDS) and its necessity in the IoT, the classification of IoT attacks, publicly available benchmark intrusion datasets and different machine learning (ML), deep learning (DL) and generative AI techniques to secure the IoT network from intrusion attacks. In this section, we also introduce an IDS model implementing some of the multiple ML and DL techniques on the publicly available CICIDS 2017 dataset. Ultimately, there is an examination of the principal obstacles that require attention from the scientific community, along with a consideration of their possible remedies.