Re: ANNOUNCE: SandMark - A Software Protection Research Tool

Stefan Monnier <monnier+comp.compilers/news/@rum.cs.yale.edu> wrote:
>>>>>> "Christian" == Christian S Collberg <collberg@cs.arizona.edu> writes:
>> * The embedder takes a Java jar-file
>> and a string (the watermark) as input and produces
>> the a new jar-file that embeds the string.
>> * The recognizer takes the watermarked
>> jar-file as input and produces the watermark
>> string as output.
>
>What do you get if you do:
>
> embedder string1 <foo.jar | embedder string2 | recognizer
>
>do you get string2 only ?


It depends on the algorithm. Most software watermarking
algorithms are keyed, so you'd say


      embedder string1 key1 < foo.jar | embedder string2 key2 > foo1.jar
      recognizer key1 foo1.jar # gets you string1
      recognizer key2 foo1.jar # gets you string2


However, even some keyed algorithms have the problem that
a second embedding is likely to destroy the first. For
example, for algorithms that embed the watermark in the
instruction frequencies of the program (e.g. Stern's
algorithm which is one of the algorithms implemented
in SandMark), multiple embeddings are unlikely to survive.


Christian


_____________________________________________________________________
            Christian Collberg | "It's wondrous, with treasures to
        christian@collberg.com | satiate desires both subtle and
          University of Arizona | gross; but it's not for the timid."

Return to the comp.compilers page.
Search the comp.compilers archives again.