Embedding Multiclass Data Hiding in Compressed Sensing by Valerio Cambareri, Mauro Mangia, Fabio Pareschi, Riccardo Rovatti,Gianluca Setti
The idea of using compressed sensing as an acquisition protocol with some encryption properties has been envisioned, but never explored in depth since its security may seem doomed by the intrinsic linearity of the encoding process. In this investigation, we quantify some aspects of straightforward statistical analysis and known-plaintext attacks showing that, although not perfectly secret, compressed sensing with universal encoding matrices grants a noteworthy level of data hiding, that may come at almost-zero cost in limited-resources applications. Moreover, the simplicity of the encoding method allows the definition of a general, lightweight scheme with which encoders may distribute the same information to receivers of different "classes", the latter being able to retrieve it with provably different quality levels. The effectiveness of this multiclass data hiding strategy is also analyzed by quantifying the chances that a low-class decoder has to reconstruct the high-class decoding keys by means of known-plaintext attacks. Examples showing the effect of multiclass encoding in a variety of settings (speech, images and electrocardiographic signals) are reported.
The simulation framework can be found at: http://securecs.googlecode.com
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.
No comments:
Post a Comment