Comprehensive Evaluation of Cybersecurity Datasets for IoT Environments

According to the vast majority of sources, assaults on systems connected through the internet of things (IoT) take place at the data transmission layer. In the field of IoT, different deep learning (DL) and machine learning (ML) approaches are utilized in order to construct the intrusion detection system (IDS) and methods that can identify attacks in one of the layers of the architecture. In this regard, the primary goal of cybersecurity could be to reduce the severity of the attacks as much as possible, while acknowledging that complete prevention is impossible. The number of people who are part of the defense and protection system is far lower than the number of persons who are responsible for the preparation of the attacks. Problems that are learning-backed and well-reasoned need to be tackled by the cyber environment, which must do so utilizing acceptable methodologies with quality datasets. The goal of this article is to analyze and summarize the various IoT assaults that were carried out in the past, as well as to discuss the enhancement of the cybersecurity datasets that have been utilized for training the techniques that were utilized for creating IDS detection methods. Additionally, the paper will detail the process by which these datasets have been employed. Datasets specifically KDDCUP99, UNSW-NB15, and NSL-KDD, is the one that is recommended by researchers the majority of the time.