An Extensive Review of Machine Learning and Deep Learning Techniques on Network Intrusion Detection for IoT

The Internet of Things (IoT) has transformed technology interactions by connecting devices and facilitating information exchange. However, IoT's interconnectivity presents significant security challenges, including network security, device vulnerabilities, data confidentiality, and authentication. Many IoT devices lack strong security measures, making them susceptible to misuse. Additionally, privacy concerns arise due to sensitive data storage. Solutions such as secure authentication, encryption, and encrypted communication are vital. Intrusion detection systems (IDS) play a crucial role in proactively protecting networks, yet they encounter significant challenges in identifying new intrusions and minimizing false alarms. To tackle these issues, researchers have developed IDS systems that leverage machine learning (ML) and deep learning (DL) techniques. This survey article not only provides an in-depth analysis of current IoT IDS but also summarizes the techniques, deployment strategies, validation methods, and datasets commonly used in the development of these systems. A thorough analysis of modern Network Intrusion Detection System (NIDS) publications is also included, which evaluates, examines, and contrasts NIDS approaches in the context of the IoT with regard to its architecture, detection methods, and validation strategies, dangers that have been addressed, and deployed algorithms setting it apart from earlier surveys that predominantly concentrate on traditional systems. We concentrate on IoT NIDS implemented by ML and DL in this survey given that learning algorithms have an excellent track record for success in security and privacy. The study, in our opinion, will be beneficial for academic and industrial research in identifying IoT dangers and problems, in implementing their own NIDS and in proposing novel innovative techniques in an IoT context while taking IoT limits into consideration.