Cyber-Attack Prediction using Dense Network Model for Intrusion Detection in Cyber-Systems

Due to the endpoint devices' computing, storage, and connectivity limitations, IoT infrastructures are susceptible to cyberattacks. One viewpoint holds that most newly created cyber-attacks are created by subtly altering existing cyber-attacks to create a new assault that the IoT network frequently ignores as ordinary traffic. The term “IoT-based Intrusion Detection and Classification System using Convolutional Neural Network” describes creating a new, autonomous, intelligent solution for classifying and detecting cybercrimes in IoT communication networks based on deep learning known as Feed-forward recurrent network model (FFRNM). The purpose of the anticipated mode is to approximate the functionality. The model allocated input to the dataset and output is fetched from the value. The suggested IoT-model uses speedy processor for parallel processing and dependable architecture for high-performance computing. Three subsystems make up the proposed system: feature learning, traffic classification, and feature engineering. Each subsystem in this research was developed, analyzed, integrated, and validated. The suggested patch was evaluated using the Network Security Laboratory-Knowledge Discovery Databases (NSL-KDD) dataset, which covers all the important dangers associated with IoT computing. According to the simulation findings, the multiclass classifier (five categories) and the binary-class classifier (normal vs. odd) had greater classification accuracy with scores of 99% and 98%, respectively.