[cgasec] Comments and clarification questions on draft-dong-savi-cga-header-03

[Tony Cheneau <[e-mail]>]

Hello,

I read through your document and I find it pretty interesting.

However, I have some comments or clarification questions:
- Section 7.1: the text references the section 3.1, I think you want to
   reference the section 5 here.
- as you noted, the attacker needs to perform more calculus that the
   node, however, you forget to mention that the SEC parameter has a big
   influence on the address generation time (for the node and for the
   attacker). Each time this parameter is incremented, the address
   generation time is multiplied by 2**16.
- clarification question on the Section 9.3: if I understand correctly,
   you need to have an exchange of 4 messages before you can trust both
   side. What happens if there is less that 4 messages ? For example if
   whole communication was initially a single UDP packet.
- also, in your document, you ask if the Signature Algorithm Agility
   work in CSI may help you. As part of the author of this work, I think
   that indeed this work may interest you.  We have results that proves
   that switching from RSA to ECC has many advantages when using CGAs
   (the advantages are usually well known, such as a shorter key size,
   quicker signature generation, etc).


I'm looking forward to have a good discussion during the bar bof.


Regards,
 	Tony Cheneau