Banking under Enduring Power of Attorney
Ian Batten
ukcrypto at chiark.greenend.org.uk
Sat, 11 Apr 2009 11:55:49 +0100
On 9 Apr 2009, at 11:39, Nicholas Bohm wrote:
>
> Do you have any details of the claimed error rates?
It's the Fujitsu PalmSecure technology (we should eat our own
dogfood), integrated into a lock system compatible with the pre-
existing equipment we have.
> Internal research by Fujitsu resulted in a false acceptance rate of
> less than 0.00007% and a false rejection rate of only 0.00004%.
> False acceptance rate is a rate at which someone other than the
> actual person is falsely recognized. False rejection rate is a rate
> at which the actual person is not recognized accurately.
>
There are two modes: one in which you load a bunch of peoples'
patterns into the system and it authenticates against the presentation
of a hand, and one in which you associate prints with proximity cards
and the system confirms that the card matches the print and the duo
are valid. We're currently running the former, but will transition to
the latter.
Clearly, once you throw a proximity card into the mix the risk drops a
little, because
(a) you have to be able to steal or duplicate the card and
(b) you have to be able to fool the system on the print that the card
is associated with, rather than an arbitrary one of the prints stored
in the system.
Or, of course, (c) you have to force recognition without the correct
credentials or a close-enough replica of them. But that's a system
attack rather than an authentication method attack, and is slightly
different to attacks on the palm print mechanism itself.
ian