Advanced intrusion detection in IoT environments using hybrid optimization and deep learning techniques

Abstract

Through integration and engagement with our surroundings via the Internet of Things (IoT), we have facilitated the emergence of more intelligent houses, grids, and urban areas. This network of gadgets enables efficient communication, allowing users to make educated decisions and optimize their daily routines. The Internet of Things (IoT) is rapidly connecting smart homes, cities, and transportation systems. However, this increase posed significant security threats. Researchers have developed novel threat detection methodologies to address these issues. This thesis proposes unique ways that integrate environmentally motivated optimization techniques with deep learning to address security concerns. These methodologies augment IoT security by enhancing the agility and durability of threat detection systems. The aim of this effort was to develop BMOHNN, a Threat Identification System (IDS) specifically designed for Internet of Things contexts. The proposed method effectively identifies and classifies threats, mitigating the security and privacy concerns related to IoT networks. To preprocess input data, identify and classify risks, and improve neural network model parameters, it integrates Hopfield Neural Network and Barnacles Mating Optimizer methodologies. The next study looks at how to use a Double Adaptive Weighting Arithmetic Optimization Algorithm along with Deep Learning (DAWAOA-DL) to create an intrusion detection system that works best in Internet of Things environments. newline