Hi Malisa,

On Oct 31, 2015, at 11:10 PM, Malisa Vucinic wrote:

> Thanks for the clarification. We are trying to maximize code reuse

Then please note that PANA is RFC, has multiple implementations, adopted by 
other standards (Zigbee and Wi-Sun), and already getting deployed (Wi-Sun).


> and minimize number of exchanges. In Q17 I see that Client Initiation message 
> is omitted if network can detect on its own the presence of JN, which is not 
> the case in 6tisch.

OK. 
Note that, though, any solution you'd have would need to have the equivalent of 
that one message where the client says "hello, let's start auth".


> I suppose that we are then left with EAP overhead plus an extra discovery 
> message.
> 

I see the EAP overhead is being debated already.
But like I said above, discovery message needs to be supported by any other 
solution as well, hence I don't think it's extra for "EAP/PANA"

Alper




> Regards,
> MaliĊĦa
> 
> 
>> On 31 Oct 2015, at 22:43, Alper Yegin <[email protected]> wrote:
>> 
>> Hello,
>> 
>> I've read the thread. I see there are some PANA-related questions. Let me 
>> clarify them.
>> 
>> Like Rafa said, PANA is already adopted and fully defined by Zigbee IP. 
>> Hence I'd expect it's ready to be used for 6tisch as well. Are there 
>> anything special here in 6tisch that'd make a difference with respect to 
>> network access authentication and key management?
>> 
>> 
>> PANA assumes the JN has an IP address. That IP address can be a link-local 
>> IP address.
>> (Not sure if this work really needs a solution that operates before *any* IP 
>> address is configured on the JN. But if it does, then here's an expired work 
>> on that front: 
>> https://tools.ietf.org/html/draft-yegin-pana-unspecified-addr-06)
>> 
>>> - Traffic load: minimum 4 messages for PANA + [ 5 messages for EAP-AKA, 15+ 
>>> messages for EAP-TLS ] + 1 packet to provision K2
>> 
>> Not sure where that "4 messages" come from. It can be "0". See Q16 and Q17 
>> in http://www.panasec.org/docs/PANA-FAQ.txt.
>> Also that last 1 packet is not needed either. The last packet sent from the 
>> authenticator can deliver that payload.
>> 
>> I don't think a separate KMP is needed when we use PANA.
>> Because:
>> - EAP/PANA authentication generates PANA SA between the JCE and JN. Which 
>> can used for further key derivation. Given the PANA is extensible thru AVPs, 
>> additional parameters can be delivered between the end-points if necessary.
>> - RFC 6786 defines how PANA can carry encrypted payloads, which can be used 
>> for delivering keys.
>> 
>> I hope these help.
>> 
>> Alper
>> 
>> 
>> 
>> 
>> _______________________________________________
>> 6tisch mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/6tisch
> 
> _______________________________________________
> 6tisch mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/6tisch

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

Reply via email to