Yubico Forum

Re: Only a few bits of the token change from each...

2008-06-04T10:47:36+01:00

This led to some questions in email so we'd like to clarify something:

This change does not change the cryptographic properties, and the change isn't done to combat any attack including theoretical attacks. We still assume AES-128 is resistant to known-plaintext attacks.

/Simon

Statistics: Posted by Simon — Wed Jun 04, 2008 10:47 am

Entropy increased

2008-06-03T23:05:02+01:00

Based on input from this forum, additional entropy is added from firmware version 1.3.0 that is due to be released soon.

In order to maintain compatibility with existing versions, we've decided to seed the 24-bit timestamp with a random value at startup rather than having it start at zero.

Jakob E
Hardware- and firmware guy @ Yubico

Statistics: Posted by Jakob — Tue Jun 03, 2008 11:05 pm

Only a few bits of the token change from each...

2008-05-20T20:40:49+01:00

Q: Only a few bits of the token change from each authentication request - and an attacker could guess some of the plaintext if they knew the approximate current sequence number - have you done any research with regard to the impact this has on cryptoanalysis?

A: We have a 16-bit random number and and a 16-bit CRC together with the counter- and timer fields. So there is a bit more stocastic means than just the counters themseleves.

We have been thinking of increasing the random part to make the string less deterministic. It would be a very simple thing to do

Statistics: Posted by hrag — Tue May 20, 2008 8:40 pm