Notes: Janfred Rieckers, Michael Richardson
Agenda remains unbashed.
TEAPv2 was just adopted, two PQC-related docs in WGLC, Update to EAP-PPT
probably comming soon, bootstrap waiting for running code, EAP-EDHOC -
Chris (AD) will probably push button soon.
Alan presenting.
Elliot: Implemented one, does not follow the draft exactly, please take
a look. Implemented X509 machinery in hostap. Like the idea of ordering.
Problem with TEAP is that its too free, idea for order (machine auth
first, then user auth, ...) is good. Signalling must be good. Thing not
mentioned yet is cryptobinding. TEAPv1 did cryptobinding to things that
weren't crypto. TEAPv2 does good job of explaining where the keys come
from. Needs to be implemented. Once code+spec is complete,
Alan: Once we have a good idea that the spec is complete, we should pull
everything referencing TEAPv1 out, so it is a standalone document.
Heikki: Says "Mandate all allowed flows"
Alan:
Elliot: We should not allow new flows in if there is not a valid use
case.
Guilin presenting.
Joe: Where do you have to make changes to implement this in the
server/backend?
Guilin:
Joe: Does the server has to keep more state?
Guilin: Yes, original only MK,K_encry,K_auth, now we also need
K_encr' and K_auth'
Joe: Do the K_encr' and K_auth' be stored persistent? even across
session?
Guilin: During session
Michaƫl and Emmanuel presenting.
Matt: I think there is a misappendence of the
Aritra: Second Matts comment, need to analyse if AES is really under
quantum thread, and is doubling the key size helpful?
Michael: AES-128 keys are not immediately threatened, but we have
regulation that mandates making keys longer.
Alan: Yes, it takes years to implement new drafts. But other question:
How widely used is EAP-PSK? Don't see much of this deployed.
Michael: Is used in a protocol using PLC communications, will look for
other use cases. Especially in PLC use cases it is necessary to be
lightweight, so EAP-PSK is helpful.
Emmanuel: Also used in many constraint devices, e.g. smart meters, that
have very limited capabilities in communication.
Heikki: wpa_supplicant supports EAP-PSK already, might be good testing
tool
Praveen presenting.
Mark: Why can't you use EAP-AKA in enterprise deployments, in
OpenRoaming it works. In UK there is Fed WiFi, which uses EAP-AKA over
third-party Enterprise network
Praveen: Openroaming uses passpoint, still requires an endpoint at the
MNO for authentication.
Mark: UK fed-wifi is not passpoint-based, that's simple EAP-AKA. Other
issue: Perhaps passpoint is not enabled, but it should be possible and
could be enabled.
Alan: Topic in radext: supplicants blasting NASes with requests,
probably include text for supplicants in the RADIUS best practices
document.