Unified multimodel fusion for precision defense against evasive denial-of-service attacks

Mitigating Denial-of-Service (DoS) attacks is critical to ensuring the security and availability of online services. Although machine learning models have been widely employed for DoS attack detection, there is a growing need for novel strategies to further enhance their performance. In this study, we propose an innovative approach based on combinatorial fusion, which integrates multiple machine learning models through advanced fusion algorithms. This approach incorporates score combinations, weighted score techniques, and leverages the diversity strength of scoring systems. Through comprehensive evaluations, we demonstrate the effectiveness of this fusion method using key performance metrics, including the number of correct predictions and recall. Our approach addresses the challenge of classifying low-profiled attacks by combining models into a comprehensive detection solution. The results of this study highlight the potential of combinatorial fusion to significantly improve DoS attack detection and strengthen overall defense mechanisms.