Encrypted Data Stream Identification Based on Pattern Recognition

In order to get the encrypted data stream type information without decrypting the encrypted data, this paper presents the identification method of encrypted data stream based on pattern recognition. First, the randomness test method NIST is used to analyze the data stream, getting the 15 kinds of randomness test values as the classification feature, then create classification models for different types of data streams; finally, using the trained model to identify the unknown data stream. Simulation results show that after optimizing the 15 kinds of features, the proposed method can effectively classify the different types of data stream, and the average recognition rate is about 85%.