Re: [6tisch] TSCH and CCM security proofs

[Tero Kivinen <[e-mail]>]

Michael Richardson writes:
> 
> Tero Kivinen <[e-mail]> wrote:
>     > Michael Richardson writes:
>     >> > Implementations MUST use different L2 keys when using different MIC
>     >> > lengths, as using same key with different MIC lengths might be
>     >> unsafe > (i.e., using same key for both MIC-32 and MIC-64). See IEEE
>     >> 802.15.4 > Annex B.4.3 for more information.
>     >> 
>     >> This seems like it isn't a problem.  It would apply to network-wide
>     >> keying only.
> 
>     > It applies all keys, not only network-wide keys.
> 
> yes, but if one uses 802.15.9, then the likelyhood of repeated keys is pretty
> low, right?

No. Broadcast keys are most likely same for some time, and if you join
the network before they are changed, the 15.9 will deliver you the
same keys than last time. Pairwise keys generated by 15.9 are of
course going to be different. 

>     > Most common use might be someone using MIC-32 for beacons, but using
>     > MIC-64 for actual data or something like that.
> 
> Yes, but why do that?
> What's the benefit?

Most likely saving bytes over the air. The EBs are sent quite often,
and things that are there are already known by anybody who is already
joined the network, thus they only care about the exact timing of the
frame, not really about the content. Of course if there is changes for
the network setup indicaed by EB, then it would be different, but I do
not think anybody does that, as chaing any of the parameters inside
the network (timing, channels used etc) would really mess up the
network completely thus tearing down the network and starting over is
better option.
-- 
[e-mail]