Re: two of three

From: Michael Brown (see_at_signature.below)
Date: 01/28/05 

Date: Sat, 29 Jan 2005 02:05:11 +1100

Bob Harris wrote:
> Suppose I have a value x (chosen from some range) and I want to
> encrypt it with three value a, b, and c (each from the same range as
> x) such that x can be recovered from any two of those values. In
> other words, we have functions f, g, and h such that f(a,b) = g(b,c)
> = h(c,a) = x.

The term to look for is "secret splitting" A primitive approach in this case
would be encrypt it with a single key, call it K, and let it be n bits long.
Then, split K into 3 equal portions, called K1, K2, K3. K1 is n/3 bits long,
etc.

The first "metavalue" a contains K1 and K2, b contains K2 and K3, and c
contains K3 and K1. From any two of these values, the original key can be
found. The downside to this approach is that you have dramatically reduced
the brute-force keyspace if you have only one of the "metavalues", something
which better methods avoid.

[...] 

-- 
Michael Brown
www.emboss.co.nz : OOS/RSI software and more :)
Add michael@ to emboss.co.nz ---+--- My inbox is always open

Relevant Pages

  • Re: FUD about CGD and GBDE
    ... single key is dependent on the block size and cipher ... the NIST specification for AES and CCM mode (NIST Special ... that leaves a huge amount of material you can encrypt with ...
    (freebsd-hackers)
  • Re: double compression considered harmful?
    ... No it would not make it less secure unless you pick a very bad key ... then encrypt that same file again with an independent key it will not ... to break it than the single key in general. ... My Crypto code ...
    (comp.compression)
  • Re: [9fans] s3venti
    ... s3venti does encrypt blocks that it writes to S3. ... uses a single key, making it rather vulnerable to dictionary attacks, ... Institut fuer Arbeitsphysiologie an der Universitaet Dortmund ...
    (comp.os.plan9)