Hey Scott,
Thanks for all the "excellent"s. Hopefully some of them are deserved :)
Couple of clarifications:
IP address --> MAC address --> Connected AP
As the MAC addresses are NOT reliable, it is a lot more straight
forward to directly map between IP addresses and APs.
In the (excellent!) experiment, you associated yourself with one or more MAC
addresses. Your IP address is more likely to change than your MAC.
You're actually both right, but neither fact really matters to the experiment.
The goal was to locate "the thing with that IP", and the network is
authoritative for which MAC address goes with that IP -- even if the MAC
address is bogus and even if the MAC address changes. In the end, all you
really care about is that the thing that is assigned a given IP address is
connected to a given access point, which bounds its location.
Directional antenna makes it imprecise.
In the (truly excellent) experiment, the goal was to find out which room
someone was in so you could find them.
Even with a directional antenna that's throwing WiFi out for a couple of
kilometers, you're still doing a lot better than the generic IP-geo service
mentioned in the draft. And part of the point of discussing ISP-based systems
is that the ISP knows how the APs are laid out and configured -- so even if
there were directional antennas, the location system could account for them.
In reality, we just did the easy thing. If we had gotten into signal
measurements, we probably could have done much better. I understand that there
are even companies that make products for these sorts of things :)
<http://www.cisco.com/en/US/prod/collateral/wireless/ps5755/ps6301/ps6386/prod_qas0900aecd8029371a.html>
In any case, the broad point is that if you know something about the
infrastructure, you can create a better location service, and that such
services are being created. I was just suggesting that it could be helpful for
the shared-addressing-issues draft to mention the impact on these systems.
--Richard
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf