Abstract

Devices associated with the Internet of Things (IoT) are rapidly becoming ubiquitous, while IoT services are expanding their reach to become increasingly pervasive. IoT devices and applications are increasingly targets of attacks as a result of their rising popularity. IoT has long been a target of cyberattacks; however, as it becomes more pervasive in everyday lives and communities, people must take this threat with greater attention. Researchers have been trying to have a complete understanding of the threats and assaults that can harm IoT networks since safeguarding the IoT is a major issue. This work discovers the use of deep learning (DL) models in several facets of cybersecurity, analysing the results and finding parallels to machine learning (ML). An ensemble of Long Short-Term Memory (LSTM), Deep Convolutional Neural Network (DCNN), and Elman Neural Network (ENN) classifiers is used by the Multi-Class Classifier with Adaptive Horse Optimization (MCC-AHO) to identify the assault types. In this work, an ensemble technique known as majority voting was utilized, which involved optimizing the weights of the many classifiers to achieve the best possible result. This proposed model for cyber-attack detection assesses their effectiveness using the most recent CIC IoT Dataset 2023, achieving an accuracy of 98.16 %.