Unknown DDoS Attack Detection Using Open-Set Recognition Technology and Fuzzy C-Means Clustering

In contemporary society, internet users face various cyber threats, including malware, phishing websites, and hacker attacks. Conventional defense software struggles to prevent unknown attack methods, making developing effective detection techniques crucial. Distributed Denial-of-Service (DDoS) attacks represent a persistent and evolving cybersecurity challenge. As such, enhancing the defense and detection of DDoS attacks is imperative. This study addresses the open set recognition (OSR) problem, a crucial pattern recognition aspect involving managing unknown categories. To address this challenge, we employ Spatial Location Constraint Prototype Loss (SPCPL) and Fuzzy C-Means methods. The findings of the experiment indicate that the unknown attack detection technique we have suggested, which relies on Fuzzy C-Means open set recognition, exhibits superior performance compared to conventional known attack detection methods in detecting and preventing unknown attacks. The misjudgment rate is extremely low given the high accuracy rate of 98% and minimal sample overlap. Finally, improving the dependability and consistency of models in real-world implementations can aid in mitigating intricate and ever-changing attack situations.