Network traffic grant classification based on 1DCNN-TCN-GRU hybrid model

Accurate grant classification of network traffic not only assists service providers in making acceptable allocations based on actual business demands, but also ensures service quality. To further improve the accuracy of traffic classification, we propose a hybrid method of 1DCNN-TCN-GRU for traffic data authorized classification. The proposed hybrid model extracts more complex features by combining the advantages of one-dimensional convolutional neural network (1DCNN), temporal convolutional network (TCN) and gated recurrent unit (GRU). 1DCNN is used to extract the spatial features of the original data. TCN is used to further extract the temporal features. GRU is used to extract the long term dependencies and classify them. We conducted experiments on two datasets Huya live data and WeChat voice data. A comparison against various other algorithms is carried out to show the superiority of the proposed hybrid approach. Therefore, four evaluation metrics, namely, accuracy, precision, recall and F1 score are chosen to assess the classification results numerically. The experimental results on two real-life datasets show that the proposed hybrid algorithm outperforms other algorithms in network traffic grant classification and has certain effectiveness.