Hi Karsten,

>> At least on my local networks this would have virtually no effect to
>> do only that.
>> The SI5 & SI5ter don't have any padding here so their content would be
>> known in anyway and they provide way enough plain text.
>
> That's a correct observation: Randomizing filling bytes would force an
> attacker to use SI messages instead, thus increasing the attack
> complexity by at least an order of magnitude due to the variability in
> the SI content and placement.

I guess, it's yet another config the operator must pay attention to.
Here, operators all use late assignement, so we first go to a SDCCH
where we have between at least 2 ciphered SACCH (so 8 bursts) at
predictable position (and it's often more than 2)

And AFAICT, their content is not variable at all (explained here :
http://lists.lists.reflextor.com/pipermail/a51/2010-July/000804.html )


> Beyond the ordering of the values, SI 5/6 messages, are filled with '00'
> bytes. My proposal would be to randomize those in the same way as the
> '2B' pattern. Even making a single byte unpredictable would increase the
> attack effort by two orders of magnitude.

What do you mean by filled with '00' ?

AFAIK those are padded with '2b' as well (and SI6 is for sure here).
It just happens that the BCCH list encoding in SI5/5ter takes up all
the 23 bytes and there is no padding. Sure there are a lot of '00' in
those encoding, but they mean something, they're not just padding and
if you change them, you change the BCCH list.


Sylvain
_______________________________________________
A51 mailing list
[email protected]
http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51

Reply via email to