Research On Fine-Grained Classification Technology Of Network Application Based On Ssl Protocol

SSL is a protocol used to ensure the security of network communication. With the widespread use of the protocol, users can browse news, read emails, watch videos, visit social networks and so on based on SSL. There are more and more encrypted network applications based on web access in the network. However, traditional traffic classification methods such as port-based and payload-based can’t meet the demand of the classification of SSL encrypted application such as SSL-based web pages and emails. So it puts forward higher requirement for network management. Therefore, fine-grained classification of the network applications based on SSL has great pratical significance.The paper puts its emphasis on studying the classification technology of encrypted applications based on SSL protocol. The main works of the paper include:(1) Proposed a kind of labeling method for the classification of SSL network applications, this paper studied the SSL protocol certificate and extension field deeply. In the SSL protocol handshake process, server_name extension field of client hello message and the certificate of server hello message provided certain identifying information. This paper made statistic and analyzed the certificate and extension field in the real network, and the corresponding relationship between the extension field and network applications as prior knowledge applied to the subsequent real-time classification research.(2)Studied the machine learning methods suitable for fine-grained classification of the SSL network applications, this paper made research on the traffic classification method deeply, and decision tree method was selected as the method in our research. In the offline environment, we used the method to classify the SSL network applications and yielded more than90%precision and recall for the fine-grained classification of particular SSL network applications during the experiment.(3)Realized the fine-grained classification system of network applications based on SSL protocol in real-time environment, in the real-time environment, we used the C4.5decision tree method, designed and realized the fine-grained classification system of the network applications based on SSL protocol in real-time environment. We also optimized the system, and for the fine-grained classification of particular SSL network applications, the experimental results showed that the system obtained more than85%classification accuracy.