ietf
[Top] [All Lists]

Re: Who is interested in wireless cards for the Adelaide IETF meeting?

2000-03-04 11:20:02
This is the same card as an Apple Airport. It is 802.11 DS, 11Mbps, and
supports Wire Equivalent Privacy (WEP). The idea here is that you need a key 
to
get on the network, but once you're on you can see all the traffic "on the
wire" that you care to. The Apple software only lets you set a 40 bit key
(actually what you do is enter a passphrase that is hashed down to 40 bits),
but I believe a 64 bit key is supported by the underlying hardware. 

[Warning...  gory crypto details enclosed.  I'm not a cryptographer,
just a protocol engineer who uses the stuff every now and then..]

I haven't found specs for the "128"-bit WEP supported by the turbo
cards, but the standard version of WEP (which I believe is what's
implemented by the Silver cards) uses RC4 with a 64 bit key but
roughly 40 bit effective strength.

RC4 is a stream cipher -- it generates a pseudo-random bitstream which
then gets XOR'ed with the plaintext.  If two (or more) packets share
the exact same keystream, there are various cryptanalytic techniques
which can be used to easily extract plaintext from one or both of
them.  (among other things, this is why reusing a one time pad is so
dangerous).  It's also usually very easy to inject chosen plaintext
into a low-level network component, which makes keystream recovery
even easier.

So, to address this within WEP, 24 bits of the 64 bit packet key are
used as an "initialization vector"; they're chosen different for each
packet and are sent in the clear in the WEP header; the remaining 40
bits are a secret shared by the nodes in the network.

However you still run into the "two time pad" problem if IV's are
reused, which will definitely occur by the time 2**24 packets have
been sent through the network using a given key (and may occur much
sooner depending on how IV's are chosen).

I hope the 128 bit "gold" cards use a longer IV..

                                                - Bill