Re: Fingerprint as cryptokey

In article <ed4u1a$clp@xxxxxxxxxxxxxxxxxxxxxxx>,
Mike Amling <nospam@xxxxxxxxxx> wrote:

However, its possible to partition an N-dimensional
space so that no point is close to more than N+1 regions.

True. I found this out myself. Where did you get it from?



The fingerprint can thus limit the generated "key" to a set whose
cardinality is linear in the number of dimensions measured. Trying all
keys in the set can't take much more time than describing the point in
N-space.

This was the idea I had long before I started at Genkey, and the
reason I took the job of developing biometric cryptography, and the
reason that I was recommended to the founder by people who knew I did
stuff like that.

I knew it could be done, and had already done a lot of work on it, so
the only thing that had to be worked out was how to analyze
fingerprints, which I already had worked partly out long before.

However, it turned out that the only solution he would accept was his
idea of one unique number from biometry, which were already patented
by Bodo in Germany. I proved that to be impossible. I told this and
that there were other ways of doing it which I had reasearched on my
own for 7 years, and he told me I lied and bragged too much.

I understood I was working for some sort of arrogant self deluded
psychopath, and made the design less general, and made it of only
publicly known parts, so that the quarantine would not stop me from
using them again after I was sacked, which I believed I would
eventually be. I was eventually sacked, despite having designed the
entire system, including the fastest ARM RSA, and being the only one
who knows how and why it actually works.

When I had started working on the fingerprint analyzer, the founder
threatened to sack me if he ever caught me doing anything not directly
ordered again. He was to direct that work. Needless to say, progress
disappeared, but I kept my job and got several years of pay by
following orders.

And partly due to my foresight, I am now free to patent and develop a
similar system, if anyone should be interested. Or perhaps I will just
make it public. Whatever.



And Mike, are you not concerned that this neighbor idea of ours could
be used to efficiently locate fingerprints of entire populations? That
is one of the reasons I have not blathered about this method.

Kim0


PS: As for the proof: The more dimensions, the higher the probability
that a point will lie on the edge between cells. For high dimensions
this become a certainty. All people sort of become 1.8 meters tall,
falling directly on the split.
.

Relevant Pages

  • Re: Java EncryptZip and DecryptZip classes
    ... >Mike Amling wrote: ... Will going to the effort to add CBC code really ... > Against what threat? ... Which isn't all that bad a reason... ...
    (sci.crypt)
  • Re: urandom sufficient for e.g. SSL?
    ... Mike Amling wrote: ... there's no reason not to use the bits you can get from it, ... if you can get enough to seed a CSPRNG. ... somewhat unfriendly to other applications on the system (who may now ...
    (sci.crypt)
Quantcast