Identification of SSH Applications Based on Convolutional Neural Network

Abstract

SSH is an encrypted communication protocol. SSH tunnel may encapsulate some other unknown applications, which has a certain potential impact on network security, so it is necessary to identify these applications accurately. This paper uses Convolutional neural network to identify applications, which has the characteristic of automatic feature learning. Therefore, traffic classification algorithm based on deep learning is used to identify these encapsulated applications in SSH traffic such as payload. Experimental methods and results are described in this paper and indicate that classification accuracy of applications (such as Nmap, Baidu Network, Netease cloud music, Netease cloud notes and etc.) in SSH tunnel is up to 95%.