Statistical Characterization of MP3 Encoders for Steganalysis: 'CHAMP3'

摘要

This report results from a contract tasking Technische Universitat Dresden (Technical University of Dresden in Germany) as follows: Grantee will investigate and characterize MP3 encoders and statistically analyze MP3 data to enable detection of embedded information in MP3 files and streaming data. This may also assist with identification of the origin of MP3 files (i.e. provide information regarding the type of MP3 encoder used to generate sample data). As detailed in the technical proposal the research consists of three parts. The first task is to survey the discipline to identify the available MP3 encoders and generate a data pool for analysis. The second task is analysis of the encoders and statistical classification of the data they produce. The third task is to characterize the robustness of techniques developed by testing the approach on known and unknown reference data.