Le 19/02/2013 18:39, Doug Barton a écrit :
On 02/19/2013 07:40 AM, Alexandru Petrescu wrote:
Such a concern would be all the more valid if there were a
specification which said 'each vehicle MUST form its IIDs based on
their VINs'. But this is not the case. We are not trying for
such a definitive document.
We are certain that some vehicles will never use this VIN-based
addressing scheme. At that point, someone receiving a packet with
a certain source address will never know whether that packet is
from a vehicle or from something else. A reversal of the IP into a
number will never be sure to be a VIN.
But the fact that _some_ of them may be VINs make it a potential
source of data mining, which should be avoided if at all possible.
Data mining should be avoided if possible. The methods proposed are not
final, and they should be improved.
I would like to add about data mining.
Outside the VIN discussion, it is already possible to not use Privacy
Addresses. For example, a recent experiment with IPv6 cellular client
shows me that the IPv6 addresses of the smartphone has the IID derived
from a manufacturer's ID.
To some extent this is a privacy issue, because data mining could find
out that the manufacturer of my equipment is that brand, and propose me
that brand when I buy a tablet on amazon.
In that sense, I doubt it is possible for anybody to impose the use of
Privacy Addresses (randon number in IID, if I am not wrong) to e.g. many
cellular operators. It is still seen as an option (I think).
The same for VIN - if a use is found about the conversion of VIN to an
IP address, it is good to build in privacy, in a certain sense. But
it's hard to make sure it's accepted.
Rejecting VIN-based addresses just because of privacy aspects is
debatable, IMHO.
There is no reason that manufacturers cannot keep their own mapping
of VIN to IID, and making this process NON-standard (that is, not
documenting a standard way to do it) will slightly decrease the value
of that data, although wouldn't necessarily help the Ferrari-nee-Fiat
problem. :)
I agree. Manufacturers may indeed keep their own mapping VIN-to-IID, or
other VIN-to-prefix mappings. This would not be visible to Internet.
I would to stress further that the two drafts we propose are simply
methods. One is to generate an IID and the other is to generate a ULA
prefix, each from same VIN.
Depending on how IID and ULA are combined, and how the connection to
Internet is made (tunnel or not, IPsec or not, etc) and how the address
architecture plan for vehicles is made, various Privacy issues may
arise, or may not.
I think I may need to actually better expose the problem: how to form
IPv6 addresses for vehicles. (yes we know these already exist: DHPCv6,
PRefix Delegation on cellular, stateless autoconf, NAT, NPT, 64share).
Alex
Doug
--------------------------------------------------------------------
IETF IPv6 working group mailing list ipv6@ietf.org Administrative
Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
