Abstract:

The Internet of Things (IoT) is one of the most rapidly growing technology today It is a technology that allows potentially billions of smart devices or objects, called "things", to collect many types of data about themselves and their environment using diverse sensors. They can then share this data for a variety of purposes, including monitoring industrial services or enhancing business services or functions. However, the Internet of Things is currently facing more serious threats from all security sectors than ever before, making it easily hackable. Today, machine learning has become a new technology to solve this problem that the Internet of Things suffers from, which opened a number of new avenues of research to solve these challenges now and in the future. However, machine learning is a super powerful technology for identifying and detecting threats, attacks and suspicious anomalous activity in smart devices and networks. In this paper, different machine learning algorithms and techniques were compared to detect and identify attacks and anomalies, in addition to determining the best possible algorithm to detect attacks based on this comparison, after reviewing the detailed literature on IoT architecture, machine learning techniques, and the importance of IoT security in the context of different types of actual potential attacks. Moreover, potential effective MLbased solutions for IoT security are presented and some challenges are discussed.