Phil Karn wrote on 1998-12-09 05:38 UTC:
> >I've always wanted to set up some secret-sharing filesystem where
> >you have to download multiple "shares" to reconstruct the data.
> >But other combinations of those exact same shares give other data.
>
> I've also been toying with this idea for a few years. Throw in
> Reed-Solomon code, and you can make a fault-tolerant network where you
> need only K servers out of N to reconstruct the data, but less than K
> are insufficient. I have written fast RS code -- it's on my website,
> http://people.qualcomm.com/karn/dsp.html.
I don't think RS is the optimal choice here, because in contrast to
forward error correction, you can arrange in this application to
let the recombination algorithm know *which* parts are missing. So you
have less degrees of freedom to deal with and can take this
into account to generate a more efficient coding algorithm, such as
Rabin's information dispersal algorithm.
M. O. Rabin: Efficient Dispersal of Information for Security, Load
Balancing, and Fault Tolerance. JACM 36(2), pp 335-348, April 1989.
Markus
--
Markus G. Kuhn, Computer Laboratory, University of Cambridge, UK
Email: mkuhn at acm.org, WWW: <http://www.cl.cam.ac.uk/~mgk25/>