DDoS Attack Detection Method Based on Improved Bagging

In view of the complexity and concealment of distributed denial of service attacks, a DDoS attack detection method based on improved Bagging ensemble learning is proposed. In order to effectively deal with the changing attack modes and improve the stability and accuracy of the detection model, this method enhances the recognition ability of DDoS attacks by introducing a weighted voting mechanism and combining multiple machine learning algorithms. Experiments on the CIC-DDoS2019 dataset show that after preprocessing steps such as feature selection, data normalization and feature correlation analysis, the improved Bagging model using RF, DT and KNN as base learners has an accuracy of $\mathbf{9 9. 9 8 \%}$ on the test set, which is significantly better than the traditional Bagging method and other detection models. This study shows that the improved Bagging method not only improves the accuracy and robustness of DDoS attack detection, but also provides new ideas for solving challenges such as high-dimensional data processing and dynamic network traffic analysis.