Identification of unknown protocol traffic based on deep learning
Ruolong Ma, Su‐Juan Qin · 2017
This paper aims at identifying unknown protocol in complex network environments, using the deep learning technique that is widely used in identification. The method identifies the protocols in the network flow according to the application layer protocol types, and find out the unknown protocols. In this paper, 200,000 traffic flow are caught, of which the payload information is regarded as a dataset. The Convolutional Neural Networks (CNN) in the deep learning is applied to train and test the new identifying model, using the Keras framework. The precision of unknown protocol traffic identification is 86.05%. The experiments validate the effectiveness of this method with high identification precision of unknown protocol traffic.