I thought I'd share some random thoughts on this proposal:

  *
First of all, I do not believe that this is a good fit for the IPsecME working 
group - it doesn't really touch on IKE at all.  On the other hand, I'm not sure 
where would be better.  The only place where this might partially fit is QIRG 
(Quantum Internet Research Group), but I'm not sure they'd be interested.
  *
I do not believe QKD networks internally run via IPv6, and so I'm not sure that 
the IPv6 extensions you suggest are appropriate.  On the other hand, I'm not 
sure how QKD networks work internally, and that's not something the IETF gets 
into - maybe ETSI would be a better fit (as they really do have a QKD group)?
  *
I believe that QKD systems don't natively generate a random value R and then 
transmit it to the other side; instead, the two sides work together to generate 
a random R that no one else knows.  For the (n,n) scheme with the xor 
combining, that is quite sufficient; we'll generate n random values R_1, R_2, 
..., R_n and xor them together.  For a Shamir (or Blakley) scheme, it doesn't 
work with random shares.  This could be worked around by having the two parties 
communicate some authenticated (but not necessarily private) auxiliary 
information; not that difficult but needs to be thought through.  Your proposal 
tries to send that information through the QKD network - I'm not sure the 
provisions for that are there.
  *
You want n distinct paths through the QKD network.  Well, one necessary (but 
not sufficient) requirement is that both peers have n direct QKD connections to 
someone else; is this realistic?  If either peer has only a single QKD link, 
there is little point to this.
  *
Another thing to worry about with any sort of protocol like this is to make 
sure that it is stable even if the peers are doing other things at the same 
time, including other instances of this protocol.  For example, if A tries to 
connect to B, B connects to C and C connects to A, we need to somehow make sure 
that each side associates the QKD shared secrets with the correct connection.  
Having two independent networks (the classical network and the QKD network) may 
make this challenging.  You try to do some of it in your QKD-DOH, but at the 
very least you'd need some sort of session identifier (to handle the case where 
there are multiple connections going on at once).


________________________________
From: 李金铭 <[email protected]>
Sent: Monday, March 16, 2026 4:13 AM
To: Tero Kivinen <[email protected]>
Cc: [email protected] <[email protected]>
Subject: [IPsec] Re: Slides for the IETF 125

hi Tero
I have uploaded the slides. I apologize for the delay.




发件人: Tero Kivinen <[email protected]>
日期: 星期一, 2026年3月16日 13:40
收件人: [email protected] <[email protected]>
抄送: Tiger Xu <[email protected]>, [email protected]
<[email protected]>
主题: Slides for the IETF 125

I am still missing two slidesets for the agenda items, i.e., ESP in
UDP for load balancing and Multi-Path Secret Sharing for QKD Key
Relay. The agenda is already very full, and if I do not receive slides
for those by the end of today, I will move them to the end of agenda
with comment (if time permits).
--
[email protected]
[DELETED ATTACHMENT IETF125-draft-xu-ipsecme-esp-in-udp-lb-15.pdf, PDF]
[DELETED ATTACHMENT IETF125-draft-xu-ipsecme-esp-in-udp-lb-15.pptx, 
Presentation]


_______________________________________________
IPsec mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to