Enhancing IoT network security through deep learning-powered Intrusion Detection System

Abstract

The rapid growth of the Internet of Things (IoT) has brought about a global concern for the security of interconnected devices and networks. This necessitates the use of efficient Intrusion Detection System (IDS) to mitigate cyber threats. Deep learning (DL) techniques provides a promising approach to effectively detect irregularities in network traffic, enhancing IoT network security and reducing cyber threats. In this paper, DL-based IDS is proposed using Feed Forward Neural Networks (FFNN), Long Short-Term Memory (LSTM), and Random Neural Networks (RandNN) to protect IoT networks from cyberattacks. Each DL model has its potential benefit as reported in this paper. For example, the FFNN can handle complex IoT network traffic patterns, while the LSTM is good in capturing long-term dependencies present in the network traffic. With its random connections and flexible dynamics, the RandNN model uses its data-learning ability to adapt and learn from network data. These algorithms boost cybersecurity by enabling defense mechanisms against challenging cyber threats and ensuring the security of sensitive data as IoT networks expand. The proposed technique exhibits superior performance when compared with the current state-of-the-art DL-IDS using the CIC-IoT22 dataset. An accuracy of 99.93 % is achieved for the FFNN model, 99.85 % for the LSTM model, and 96.42 % for the RandNN model in detecting. Moreover, the models have the potential to enhance intrusion detection in IoT networks by generating swift responses to security problems in IoT networks.