From: Kate Everitt (everitt@cs.washington.edu)
Date: Wed Nov 24 2004 - 07:40:14 PST
This paper challenges the security of the 802.11 protocol. They discuss
several hacks that compliant installations will likely fall prey to. One
key insight is that with wireless media, transmission interception is a
lot easier than wired because anyone with a radio can overhear the
traffic. There are 3 main types of hacks discussed. Keystream reuse is a
problem because two messages under the same IV will let the attackers get
the plaintext of one message if they have the plaintext of the other
message, or create a decryption dictionary by leraning the value of the keystream. The
message authentication scheme also has a problem, as CRCs are not designed
to detect tampering, but rather random errors, so they can be adjusted
with the difference of the message and the message changed, or
recalculated to authenticate a new, fake message. There is also the
challenge of authentication spoofing.
This paper is very relevant to today because this standard is still in use
(with increased keys, but that doesn't solve these problems.) It is
actually growing in popularity because it is so easy to set up, for both
businesses and personal use. The authors also present good
recommendations, from specific ideas about how the standard should have
been designed, to the comment that protocols should be released to the
security community before the are adopted. I was very surprised that it
were not, because this is a very commonly used protocol. The one thing
they didn't mention, probably because it was more of a usage scenario than
anything, was the fact that a lot of home users don't even use WEP
encryption. For any standard, there has to be a good usage model or it is
totally useless. Many users do not know this, but most should know that
end to end security such as ssh is necessary. Internet explorer even tells
you your traffic is visible. The authors could have made the paper better
by proposing a solution to WEP that specifically dealt with the issues
they discussed and mentioning that the usage model needs to be taken into
account when implementing such a solution.
This archive was generated by hypermail 2.1.6 : Wed Nov 24 2004 - 07:40:15 PST