Improving Network Security with Hybrid Model for DDoS Attack Detection

One of the most devastating forms of attack in the modern day is cyber-attacks. DDoS (Distributed Denial of Service) attacks are one kind of cyberattack. It is a kind of cyberattack in which the perpetrator takes over a network or a system and disables it, either temporarily or permanently, preventing its intended users from accessing it. Put simply, it's an attack where the target computer is bombarded with needless requests in an effort to overload it, crash it, and prevent people from accessing that network or machine. So, the most important thing to do to guard against DDoS attacks is to separate network data as soon as possible. In this study, we provide a new approach to intrusion detection algorithms known as CNN-GRU Attention. It combines Gated Recurrent Unit (GRU), Convolutional Neural Network (CNN), & a self-attention mechanism to improve algorithm accuracy. The model's structure is intended to effectively represent the geographical and temporal aspects of network flows. The experimental findings highlight the effectiveness of deep learning in cybersecurity applications, outperforming standard models. The experimental outcome showed that suggested hybrid model outperformed the competition on CICDDoS2019 data set, suggesting that it might be useful for detecting and categorising DDoS attack traffic. We evaluated the proposed system compared to state-of-the-art models. We are pleased to report that the accuracy of our work was 99.6%.