A novel optimized deep learning based intrusion detection framework for an IoT networks

The burgeoning importance of Internet of Things (IoT) and its diverse applications have sparked significant interest in study circles. The inherent diversity within IoT networks renders them suitable for a myriad of real-time applications, firmly embedding them into the fabric of daily life. While IoT devices streamline various activities, their susceptibility to security threats is a glaring concern. Current inadequacies in security measures render IoT networks vulnerable, presenting an enticing target for attackers. This study suggests a novel dealing to address this challenge through the execution of Intrusion Detection Systems (IDS) leveraging superior deep learning models. Inspired by the benefits of Long Short Term Memory (LSTM), we introduce the Genetic Bee LSTM(GBLSTM) networks for the development of intelligent IDS capable of detecting a wide range of cyber-attacks targeting IoT area. The methodology comprises four key execution: (i) collection of unit for profiling normal IoT device behavior, (ii) Identification of malicious devices during an attack, (iii) Prediction of attack types implemented in the network. Intensive experimentations of the suggested IDS are conducted using various validation methods and prominent metrics across different IoT threat scenarios. Moreover, comprehensive experiments are conducted to evaluate the suggested models alongside existing learning models. The results demonstrate that the GBLSTM-models outperform other intellectual models in terms of accuracy, precision, and recall, underscoring their efficacy in securing IoT networks.