Design of DDoS Attack Detection Model Based on DBN-DNN

DDOS attacks can occur in any system that supports internet communication, and with the continuous development of network devices and technologies, the demand for rapid and effective identification and response to attacks is also increasing. When attackers exploit 0-day vulnerabilities to launch attacks, the targeted systems can rely on deep neural networks (DNN) and deep belief networks (DBN) and other deep learning models to detect the attacks. The article first transforms the attack detection problem into a classification problem, integrating deep learning algorithm frameworks to detect target traffic. Then, by comparing the detection performance of DBN and MLP models, it evaluates a comprehensive detection model, the DBN-DNN model, which fully utilizes the excellent feature extraction and classification characteristics of the two models. Finally, through experiments, a predictive accuracy of up to 94% is achieved, demonstrating that the optimized model exhibits good performance in DDOS detection.