GSM and A5

[ifarqhar@macadam.mpce.mq.edu.au (Ian Farquhar)]

I last week had a chat with someone at Austel, mainly to try and
chase down the full protocol specifications for AMPS, GSM and pagers.
During the conversation, the subject of A5 was raised.  Here is a
quick summary of what was said:

1. A5 and A5X are no longer the current names of the algorithms.  Here
   are the new names:

	A5 is now A5/1 (purportedly "military grade" crypto)
	A5X is now A5/2 ("we export to anyone" crypto)

   A5/2 is unsupported by any existing equipment, and there is a LOT
   of discontent within the international Telco community from the
   countries who are being told they will be given A5/2.  Personally,
   I rather sympathise with them.

   Note that A5/[12] is only ever used between the GSM phone and the
   local cell station anyway, and once it gets to the carrier's network
   it's in the clear anyway.  This is public information.

2. The argument reported by the CPSR (and others) between the Australian
   (ASIO, AFP etc) who wanted A5/2 and the telcos who wanted A5/1 did
   not, according to the person I spoke to, bear much relation to reality.
   All three Telcos are using A5/1 (although Vodophone, when I rang them
   to ask which they were using called this information "classified".
   I will not comment. :)  Unfortunately, everyone is keeping VERY
   quiet about what did happen.  Time to speak to my local member of
   parliament, I think.

3. (HERE'S THE BIG ONE).  The description of the A5 algorithm which was
   released to the network was a VERY early design, and I am told bears
   little relationship to A5/1 as implemented now in GSM.

Just thought people might be interested...

							Ian.