Re: [core] [ALU] Re: Question reg. draft-fossati-tls-iot-optimizations-00

["Fossati, Thomas (Nokia - GB)" <[e-mail]>]

Hi Simon,

On 02/11/2016 16:08, "core on behalf of Simon Bernard"
<[e-mail] on behalf of [e-mail]> wrote:
>Hi all,
>
>    I'm not sure to understand the point here.
>    *   The "draft-fossati-tls-iot-optimizations-00" says "The privacy
>issue associated with the use of a long-term identifier
>    must be taken into consideration."
>    *   Thomas says "I think privacy preservation should be a goal".

>    I would like to understand which privacy concern we would like to
>achieve exactly ? With TLS we have end to end encryption. You want to
>add a kind of  anonymity ? or maybe protect ourself from connectionid
>spoofing ?

The issue is that something like CID will make tracking a device activity
across different transports very easy.  (Personally, I think NAT rebinding
is a bit different because it usually happens without client awareness,
and therefore already exposes correlation information to a possible
tracker.)

In any case, if we make this an extension of the general TLS protocol we
need to make sure we design it in a way that a) is fit for the purpose
from a functional and security perspective, and b) takes into
consideration the tracking aspects for clients that want to have finer
control on their privacy.


>    From my point of view, the connection id is just a way to replace
>the IP address by a connection identifier for use-cases where IP address
>is not fixed. So we have the same security level with connection id than
>fixed IP. We are maybe a bit more exposed to spoofing as connectionid
>spoofing is probably more simple than UDP IP address spoofing, but not
>so much. I mean connectionid is just another way to retrieve security
>context needed to decrypt Application Data.

ISTM that if CID has enough randomness and is integrity-protected, then
spoofing is not an issue (or it is less an issue than security context
lookup based on a fully tamperable 5-tuple).  But certainly this is a
dimension to explore further (e.g. depending on the way CID is
synchronised on the two sides, there might be opportunities for an
attacker that can selectively drop packets from the network to do
different things, I guess).

Cheers, t