Scott Craver
Department of Electrical and Computer Engineering
Binghamton University
ES building room 2310, phone (cell) 607-727-7166
  • Example Course:
    • EE 405, Cryptography and Information Security
  • Links:
    • Binghamton University
    • Princeton University
    • Northern Illinois University
  • Contests
    The Underhanded C Contest Watson School Trivia questions (with answers)
  • About Me:
    My CV (pdf)
    Teaching Statement
    Research Statement

  • Oh cool, I have an Erdős Number of 3

    I previously thought that my Erdős number was 4, since I published with Boon-Lock Yeo, who published with Robert Calderbank, who published with Ron Graham who published with Erdős. I now find out that my number is 3, because of a conference paper co-authored with Julien Stern. Stern and his colleague Julien Boeuf reverse-engineered and broke one of the candidate SDMI technologies for audio watermarking; in particular, he reverse-engineered a technology that my team did not. He and I wrote a conference paper summarizing the lessons learned from our two teams. Anyway, Julien Stern has an Erdős number of 2, on account of publishing with Noga Alon, Michael Krivelevich and Andrew M. Odlyzko, each of whom has an Erdős number of 1.

  • Example publications

    I am accumulating my existing publications, some of which are hard to track down in electronic format. Here are a few of the major works, below.
  • Here is a link to a post summarizing our Supraliminal Channel in a Videoconferencing Application. This has appeared in the 10th Information Hiding Workshop; a further analysis of the data rates can be found here; they show the startling result that if a supraliminal channel admits even a tiny error rate, a warden can use it to derail key exchange.

    An attempt to extend the idea to audio channels is here.

  • Watermark Attacks
    1. Can Invisible Watermarks Resolve Rightful Ownerships? Introduces the ambiguity attack, also called the inversion attack, the Craver attack, or the "IBM attack" (the people at IBM call it "The RSS attack," and I assume the extended version would be "RSS/E.")

      This is the original IBM research report, later published in proceedings of SPIE.

    2. Resolving Rightful Ownerships with Invisible Watermarking Techniques. This is the journal version, expanded with an attack on our old fix, and a better, correct fix. I actually point people to the first paper because it has less verbiage and is more direct about the big idea.
    3. On the Invertibility of Invisible Watermarking Techniques.
    4. Our Article in Communications of the ACM. This cannot be downloaded without an ACM portal account.
    5. Reverse-engineering a watermark detector using an oracle. This outlines how we broke the first BOWS watermark, and the "snakes on a cone" attack we developed to reverse-engineer watermark parameters.
  • SDMI and other real-world attacks
    1. Reading Between the Lines: Lessons Learned from the SDMI Challenge. This is the paper that got us into a legal battle with the RIAA.
    2. Analysis of Attacks on SDMI This summarizes the watermarking attacks from the challenge.
  • Protocols
    1. Zero-knowledge Watermark Detection. Proposes several early approaches to applying zero-knowledge proofs to watermark detection.
    2. Copyright Protection Protocols based on Asymmetric Watermarking. S. Craver and Stefan Katzenbeisser.
    3. Security Analysis of Public-key Watermarking Schemes, S. Craver and Stefan Katzenbeisser.
    4. On Public-key Steganography in the Presence of an Active Warden. This is a solution I presented in 1998 to steganographic public-key exchange in a hostile environment. This is a hard problem because the participants are not allowed to use crypto, or otherwise exchange suspicious data, like random-looking strings.
  • Other Things
    1. Multilinearization Data Structure for Image Browsing This introduces our PHOOSball data structure for efficient storage and search of high-dimensional datasets. This was patented by Intel Corporation, whose patent awards helped me survive my first year as a Ph.D. student.
    2. Patents! All of these were filed with co-conspirators Boon-Lock Yeo and Minerva Yeung, at Intel corporation. They are based on the above multilinearization data structure research.
      1. Multi-linearization data structure for image browsing US Patent 6,233,367.
      2. Displaying ordered images based on a linearized data structure US Patent 6,556,723.
      3. Creating a linearized data structure for ordering images based on their attributes US Patent 6,628,846.
      4. Linearized data structure ordering images based on their attributes US Patent 7,016,553.
Created with vi