Enhancing DNS Attack Classification with Convolutional Neural Networks and Gated Recurrent Units

Enhancing DNS Attack Classification with Convolutional Neural Networks and Gated Recurrent Units

연구 분야: Artificial Intelligence

학회: 2024 15th International Conference on Computing Communication and Networking Technologies (ICCCNT)

DNS attacks can be caused by various methods such as Rebinding attacks, Cryptojacking and Domain Generation Algorithm. Classifying DNS attacks in a timely manner and taking necessary action to mitigate the issue is important. This research study describes the hybrid model of Convolutional Neural Network (CNN) and Gated Recurrent Units (GRUs) that is used to improve the detection or categorization of DNS attacks. A dataset called CIRA-CIC-DoHBrw-2020 is used to train the suggested hybrid mode. According to experimental data, the LSTM prototype outperformed CNN with respect to identifying or categorizing DNS attacks. The CNN-LSTM and CNN-GRU hybrid models have the ability to improve the model’s accuracy. CNN was used for feature extraction and LSTM for temporal modeling in the CNN-LSTM hybrid model. The suggested CNN-GRU hybrid model, on the other hand, obtains the greatest accuracy of 95.16%, demonstrating the efficacy of integrating GRU with CNN to effectively recognize or categorize both temporal and geographical characteristics of DNS network data. The CNN-GRU model’s enhanced performance demonstrates how well merging different neural network architectures-like CNNs and GRUs-may enhance convergence and lower error rates.