Designing an Automated DDoS Attack Detection Framework using Attention and Nested Dilation-based TCN

Information Technology (IT) provides flexible, scalable, and virtualized resources to users on demand, helps to minimize maintenance efforts and less infrastructure expenses. These services are delivered over the Internet, managed by various administrative entities, and rely on standardized networking protocols and formats. However, the core technologies and outdated protocols may have flaws and security weaknesses that could be exploited by attackers. One of the most prevalent and damaging threats is the Distributed Denial of Service (DDoS) attack. In such an attack, cybercriminals often exploit known or unknown system vulnerabilities to take control of multiple unsuspecting computers. To address these issues, many researchers have turned to machine learning, a branch of AI that helps systems to learn from data without explicit programming. While machine learning holds promise in detecting cyber attacks, most conventional machine learning techniques struggle to handle the vast volumes of data involved in these attacks, primarily because it relies on relatively simple learning processes. Hence, an efficacious detection method is necessary to protect online services from attackers. This work presents a novel deep learning framework for detecting DDoS attack. Initially, the required information is gathered from the benchmark data resource. Further, the collected information is directly applied to the detection phase. Here, the DDoS attack detection is performed by utilizing the designed approach called Attention and Nested Dilation based Temporal Convolutional Networks (AND-TCN). This designed model effectively learns the pattern and performs effective DDoS attack detection. The experimental outcomes proved that the designed framework outperformed than the conventional methods in terms of performance.