Intrusion Detection in Wireless Sensor Networks using Hybrid Deep Belief Networks and Harris Hawks Optimizer

A substantial amount of nodes equipped with sensors make up a wireless sensor network that collects and sends data to a base station. The wireless sensor networks, subsequently, face a number of security vulnerabilities as a result of shortages of resources, installation methods, and routes of communication. Therefore, detecting illegal entry is crucial to enhancing the security aspects of wireless sensor networks. Systems for detection of network breaches offer these functions, making them necessary for every communication network. Although machine learning (ML) techniques are often utilized in intrusion detection systems, their effectiveness when managing unbalanced assaults is unsatisfactory. This research work developed an attack identification system based on deep belief networks to address this issue and boost performance. The best characteristics from the dataset are chosen using a Harris Hawk optimization technique, and the parameters that were chosen are then employed as the building blocks of a deep belief network’s structure to look for breaches. The experimental findings shown that the suggested DBN effectively detects assaults and outperforms conventional deep learning models like the Convolutional Neural Networks, Long Short Term Memory Networks, Generative Adversarial Networks and Radial Basis Function Networks.