<![CDATA[Distributed Denial of Service (DDoS) attacks pose a serious threat to defense network infrastructure by disrupting communication services that are critical for command-and-control operations. Existing detection methods often struggle to simultaneously capture the spatial characteristics of network traffic and the temporal evolution of attack patterns, limiting their effectiveness against diverse DDoS attacks. This study proposes a hybrid CNN-GRU model that integrates one-dimensional convolutional neural networks for local feature extraction with gated recurrent units for temporal dependency modeling, enabling more comprehensive representation learning than single-model approaches. The proposed framework was evaluated on a five-class subset of the CIC-DDoS2019 dataset containing 225,000 network flow records and compared with standalone CNN, GRU, and LSTM models under identical experimental settings. Experimental results demonstrate that the proposed CNN-GRU achieved the best performance, obtaining an accuracy of 0.9707 and a macro-F1 score of 0.9707, consistently outperforming all baseline models. The novelty of this study lies in the effective integration of complementary spatial and temporal learning mechanisms for multiclass DDoS attack detection, providing a more robust classification framework for defense-oriented network traffic analysis. These findings indicate that the proposed model offers a practical solution for intelligent intrusion detection and early warning systems, supporting resilient and secure defense network operations against evolving DDoS threats. Future work will extend the model to the complete twelve-class attack taxonomy and investigate attention mechanisms to further improve discrimination among closely related attack categories.]]>
