Hybrid Traffic Identification Based on Sequential Pattern Mining

A novel identification method based on sequential pattern mining is proposed to detect different network traffic in encrypted hybrid data. After dimensionality reduction for traffic series, the data are cut into short sequences. By sequential pattern mining and pruning, the characteristics sequential sets, which are extracted from transaction database comprised of these sequences, are used to calculated and analyzed the identification results. With removing each traffic's characteristic away from the hybrid data, the final results are verified and concluded. The experiments show that the recall and precision of identification results are more than 60%.