(Choosing "not ready" only as this is an early review.)

I think the authors are correct that this document doesn't create
interesting new security issues. (I'm not clear if there's really
much experience with DLEP security though, so consider that a
tentative opinion.)

I do wonder if the efficacy of the traffic classification mentioned
here would be adversely affected by that traffic being encrypted.
It's fine that the authors consider that as the work proceeds.