[KEYPROV] Privacy Enabled Provisioning

[Anders Rundgren <[e-mail]>]

F.Y.I.
As some of you know I'm since way back working with a "TPMish" token and
associated provisioning protocol that is intended to support mass-markets like:
http://www.google.com/wallet

There are strong motives for innovation in this space since the state of credential
provisioning is technically at least 10 years behind the rest of mobile/PC ecosystem:
http://webpki.org/papers/keygen2/KG-vs-KG2.pdf

Anyway, the starting point was addressing the high-value market where strong
authentication is already in use and where privacy aspects (w.r.t. to device identity
NB), are secondary.  In fact, the original E2ES (End-to-End Security) scheme tries
to emulate the methods used for card issuance while moving the "production"
out on the web.

Since the most important thing of all for a standard-to-be, namely adoption,
could be hampered by a scheme leaking device identity (during provisioning),
I have recently upgraded the system to support a fully anonymous provisioning
mode.  Because anonymous relations probably do not constitute of high-value,
elaborate methods like DAA (Direct Anonymous Attestation), were (after careful
consideration), rejected as they could stifle acceptance in spite of their merits.
It seems that anonymity has more applications in three-party relations like
payments and "I'm over 18" scenarios.

In the revised scheme the issuer requests a specific mode and it is the user
that in the case of E2ES mode will have to decide if he/she wants to share device
identity with the issuer.

Somewhat surprising the PEP (Privacy Enabled Provisioning) mode affected less
than 1% of the code in the token as well as in the protocol.

- Anders
http://webpki.org/auth-token-4-the-cloud.html