A Hybrid deep learning-based approach to detecting cyberattacks in the Internet of Things (IoT) environment

Abstract

Abstract

Due to the increasing complexity of cyber attacks and rapid growth of the Internet of Things (IOT) network, the study suggests a hybrid deep learning framework to detect cyber attack in the IOT network. To properly analyze both structural and temporary patterns in network traffic, the suggested approach is a mixture of deep nerve network (DNN) and long short -term memory (LSTM) network. Several attacks, including distributed Daniel-Off-Services (DDOS), injection attacks and data leakage, were embedded in large ton-IOT datasets that were used to measure models. The false alarm rate of the hybrid model was very low and provided an unprecedented address accuracy of 100%. Its excellent performance can be replaced by model capacity to dynamically optimize with changing attack techniques, learn deep non -lineer features through DNN layers, and analyze complex temporary sequences through LSTM block. With strong scalability abilities and large data and future threats, these findings describe an encouraging path towards promoting IOT security for industrial and mission-mating uses.