Hybrid Deep Learning Techniques for Improved Anomaly Detection in IoT Environments

Abstract

The Internet of Things (IoT) is no longer limited to single personalities, but rather, it is a perceptions that has widely increased and spread in some applications or fields. The mechanism for communicating between IoT devices similarity works as traditional communication between hosts. However, the growing use of IoT has been gaining the interest of a growing number of attackers. Hence, a number of researchers are attempting to build an intrusion detection system utilizing machine learning and deep learning algorithms. In this work, a novel attack detection model is proposed by superimposing Whale Optimization Algorithm and Bidirectional Long Short-Term Memory (WB-LSTM) together. There are numerous deep learning competencies, but LSTM is one of the ones used to interpret big data or time series data. But, it is not easy to find what is the best weights for LSTMs in order to directly achieve performance. The LSTM results were 99.1%. Hence, in this work, we introduce the WOA-LSTM hybrid model, that utilizes WOA for finding the optimal weights for a network based on LSTM, and is used to detect the IoT attacks. The 99.98% was obtained from the WOA-LSTM hybrid model.

Keywords

• CICIoT2023 Dataset, Feature scaling, IoT Security, Deep learning, LSTM Models, Whale optimization algorithm (WOA), Detect IoT Attacks.