On Wed, Aug 21, 2019 at 08:18:03AM +0000, Pascal Thubert (pthubert) wrote:
> Hello Michael
> 
> I agree that the details of how it is done in practice belong to minimal 
> security.
> My expectation would be that we discuss times when it is appropriate to 
> rekey, and what it takes to do that.
> 
> Out of my hat (but please come back with cases that I missed) I can see that:
>  
> - we need to rekey to expel undesired nodes.
> - we need to rekey if a short address is reassigned to avoid nonce-replay 
> attacks with an ASN in the past
> - the ASN-based nonce never wraps in practice, but should we reset ASN -or 
> allow it to go back in time - for whatever reason, we'd need to rekey as well.
> - based on Mirja's comment - seconded by Benjamin - minimal security should 
> be a normative reference since it expands on the security considerations
> 
> I think it does not hurt to have a word on that in the architecture, even if 
> more details are found in minimal security

That basically matches up with what I was thinking.
(And thanks to those who pointed out 802.15.9; I had forgotten that was an
option, albeit not necessarily a good one for all cases.)

-Ben

_______________________________________________
6tisch mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6tisch

Reply via email to