Smart LSTM-based IDS for Heterogeneous IoT (HetIoT)

Distributed denial-of-service (DDoS) is the most formidable attack, and many researchers are focusing their attention on safeguarding the heterogeneous IoT (HetIoT) networks from it. The paper proposes a smart intrusion detection system (IDS) using long-short-term memory (LSTM), i.e., smart LSTM-based IDS for the HetIoT. The paper highlighted data preprocessing steps after thorough exploratory data analysis (EDA) and feature extraction using the principal component analysis (PCA) technique. The proposed smart LSTM-based IDS successfully identifies and mitigates the various DDoS attacks. The paper considers binary and multi-class classification (7-class and 13-class) of DDoS attacks for efficient detection of attacks. The performance of the proposed smart LSTM-based IDS is compared with two state-of-the-art deep learning approaches, and the results reveal that the proposed smart LSTM-based IDS outperforms it. The proposed LSTM-based IDS successfully identifies various DDoS attacks with an accuracy rate of 99.98% for binary classification, 98% for 7-classes, and 99.97% for 13-classes. The work also compares the individual accuracy of 7-classes and 13-classes with state-of-the-art work. Also, the proposed smart LSTM-based IDS is lightweight, simple, and less complex than the existing state-of-the-art work models.