DDoS Attack Prediction Method Based on ITN-DTA

Aiming at the inability of traditional DDoS detection methods to identify potential attacks in advance, this paper proposes a DDoS(Distributed Denial of Service) attack prediction method based on ITN-DTA (Improved TimesNet and Dynamic Threshold Analysis). By analyzing network traffic, attacks are predicted in advance to minimize their impact. The experimental results show that compared with mainstream network traffic prediction methods, ITN-DTA achieves the lowest mean absolute error (MAE) and mean squared error (MSE), in four datasets. Compared with the fixed threshold identification method, ITN-DTA improves the prediction accuracy by 5.4%, reduces the false positive rate (FPR) by 13.9%, and reduces the false negative rate (FNR) by 6.6%.