Detect Denial of Service Attack using Hybrid Deep Neural Network

Abstract

Network hacking has become more resource-intensive in recent years, particularly in firms and organizations that rely on the internet, such as Amazon, because the hacker's goal is to prohibit the user from using the network's resources, whether the target is offensive or pecuniary. As a result, there are multiple methods for detecting intrusion and network penetration that use various ways to distinguish between legitimate and unauthorized users, including artificial intelligence systems (AI), deep learning (DL), and machine learning (ML).Organizations' network infrastructure is vulnerable to several threats, including break-ins, security breaches, and system exploitation. A network's Network Intrusion Detection System (NIDS) detects such penetration attacks and intrusions within the network. Pattern matching can easily detect recognized attacks, but new attacks are more difficult to detect.This study used a deep learning technique to build an intrusion detection model.Deep learning can perform better than classical machine learning used in earlier works in extracting features of enormous data while considering the vast cyber traffic in real life. In conclusion, the suggested approach involved training an IDS model using a hybrid deep neural network using the entire NSL-KDD Dataset.The experimental results obtained from the system approved its success within a 100% accuracy value, which is considered a perfect detection rate of four attacks(DoS, Probe, U2R, and R2L), and for detecting DoS, the accuracy was 99% in NSL-KDD.