AEFETA: Encrypted traffic classification framework based on self-learning of feature
Jingru Yang, Yuanbo Guo · 2021
This paper proposes an end-to-end framework based on feature self-learning, AEFETA, to handle the task of encrypted network traffic classification. In the AEFETA, we propose a data preprocessing method based on the structure information of the network traffic, and at the same time, a lightweight deep learning model combined with an attention mechanism also be proposed in this framework. We can convert the raw pcap file into a file with the data type that can be input to the deep neural network through data preprocessing, and then, automatically extract the spatial-temporal features perform classification tasks. AEFETA framework achieved a recall of 0.98 and precision of 0.97 in the encrypted traffic classification task, meanwhile, we find the best data format through experimental comparison.