A Mobile Application Recognition Method Based on Deep Learning Sequence Analysis

Abstract

The invention discloses a mobile application identification method based on deep learning sequence analysis. The method includes the following steps: collecting data packets generated by mobile terminals connected to a local device and forwarding the data packets to a server through a data acquisition module located at an emission source AP end of a wireless local area network; identifying TCP protocol data streams classified according to the terminals from the data packets that are forwarded to the server and received from multiple protocols of different terminals; inputting each data packetin the TCP protocol data streams to a deep belief network for feature extraction and data dimension reduction, and obtaining multiple processed TCP data packets; classifying and processing the TCP data packets by using the deep belief network and an LSTM type RNN network to obtain identification results of the LSTM output; and adopting an application identification module to match terminal information with the application identification results in a one-to-many mode, and outputting a matching result in an intuitive way. The scheme of the invention fully utilizes the advantages of deep learningin sequence analysis, and combines the characteristics of mobile application data to achieve accurate application identification under the premise of protecting user privacy.

Description

Technical field [0001] The invention relates to a mobile application identification method based on deep learning sequence analysis. Background technique [0002] Application identification refers to distinguishing different applications carried on the same type of application protocol based on the characteristics of the application itself; mobile application identification specifically refers to application identification of applications on mobile terminals, such as smart phones. With the development of the mobile Internet and the popularization of smart phones, tens of thousands of mobile phone applications covering all walks of life have been born on the mobile phone platform; at the same time, Wi-Fi coverage has increased substantially, and more and more public places have been installed With the addition of public Wi-Fi, users have greatly reduced the time and space restrictions on using mobile phones. In order to better understand the preferences and needs of users and hel...

AI Technical Summary

Problems solved by technology

[0007] For deep packet inspection, a large amount of manpower is required to build a feature library, and the user's privacy information may be violated in the process of obtaining keywords, and the encrypted data will lose its effect; for other machine learning methods, due to the use of features The information is summarized manually, so it is impossible to distinguish application information at a finer-grained level, and it often only achieves results in traffic classification

Images