Intelligent LSTM (iLSTM)-Security Model for HetIoT

Distributed denial-of-service (DDoS) is the most recent lethal threat, and several industrial and academic researchers are concentrating on defending the heterogeneous IoT (HetIoT) infrastructure from it. The research presents a novel intelligent security system using deep learning (DL)-based long short-term memory (LSTM) techniques, i.e., the iLSTM-Security model, for the HetIoT network. The research addressed the steps needed to prepare the data after complete data analysis and feature extraction using the principal component analysis (PCA) method. The research also highlighted the asymptotic time complexity analysis for the proposed iLSTM-Security model. The proposed iLSTM-Security model efficiently identifies and nullifies the different DDoS threats. The research analyzes binary (2 class) and multiclass classification (7 class and 13 class) for optimal DDoS threat detection. The proposed iLSTM-Security model’s efficacy is assessed against two state-of-the-art DL approaches, and the findings show that the proposed iLSTM-Security model surpasses them. The proposed iLSTM-Security model effectively recognizes different DDoS threats with an accuracy rate of 99.98% for 2 classes, 98.8% for 7 classes, and 99.97% for 13 class classifications. Additionally, the research assesses the individual accuracy of 7 classes and 13 classes with state-of-the-artwork. Further, the research reveals that the proposed iLSTM-Security model is lighter, simpler, and considerably less complicated than the existing state-of-the-art models.