On Fri, 11 Feb 2022 09:07:17 +0000, Andy Burns wrote:
>> b. Wi-Fi can easily be set to_not_ report_any_ nearby access points
>
> can it?
Hi Andy,
I hadn't expected anyone to even comprehend the topic, let alone ask
questions, although your question isn't about the cellular tower tracking.
As for Wi-Fi tracking, I believe the answer is yes, and, better yet, I
believe I have personally implemented that "Yes" (much unlike a nospam yes).
> even with wifi "off" the phone can still use the radios for location,
> rather than for communication, there used to be a setting to make "off" mean
> "really off" but I can no longer see it, and why should you trust it?
I believe I turned off wi-fi tracking on my phone, but what you & I mean by
wi-fi tracking may or may not be the same thing since it's multiple steps.
Bear in mind that you're probably not even thinking of some of the things
I've disabled already, such as having my own phone not shout out all the
hidden AP's that are set inside of it when I'm away from home (which is no
small feat as you have to understand the process to understand that fix).
I suspect you are aware there are _hidden_ uploads of wifi information, incl
a. The time of day on the phone;
b. the location of the phone (usually via gps but maybe not always);
c. the signal strength of all visible access points (i.e., not hidden);
d. the bssid of those visible (not hidden) access points;
e. the ssid of those visible (not hidden) access points;
f. And some other information which is stored up the uploading log.
If you're asking about that, I have taken steps to turn all that off, Andy.
Is _that_ what you're asking about?
Because I've also set up my home access points not only with _nomap (which
doesn't do anything with respect to uploads) but also to be "hidden" (which,
as we all know, isn't for security but for privacy since they're ignored).
BTW, long ago, Google was caught uploading that information even if you had
it turned off, but Google said that was a bug which they long ago fixed.
Also they used to collect all visible access points (again, not hidden) via
the Google Electronics Car but they also said they stopped that practice.
>> c. Cellular geolocation is highly unlikely by apps
>
> true but "they" can do it from the network side.
Well, maybe it's not true. I don't know.
It was when nospam claimed it was likely that I opened this thread.
I don't care if it's likely or not, nor do I care if nospam is wrong (he
almost always is wrong); but I do care about learning about privacy.
In the PDF I posted (see URL in sig), the abstract says these 5 things:
a. To triangulate cell towers, you need an accurate cell tower location
b. There is no accurate cell tower location lookup available to the public
c. However, wardriving data allows an "estimate" of cell tower locations
d. If you're not careful, wardriving methods are highly inaccurate
e. But if you're careful, and if you limit the scope, they can be accurate
Notice it's a chicken-and-the-egg scenario.
You need cell tower locations in order to use them to triangulate location.
The authors found that the public sources of cell tower locations sucked.
They found that even AP location errors were on the order of 40 meters.
They said celltower localization was worse due to greater system complexity.
They obtained the actual location of 54 Los Angeles cell towers.
Applying naive algorithms to a wardrive trace resulted in 40km errors! (WTF)
They found that the only hope you have is for celltowers bounded by a trace.
Their restricted algorithm reduced the localization errors by one half.
They collected three traces between February & March of 2009 by wardrivers.
a. Downtown LA
b. Residential LA
c. Rural Victor Valley in San Bernardino County
At 2 second intervals at 32kmph they collected 2,613,465 RS readings from
105,271 uniqu3e locations, averaging 25 RSS readings per location.
They were aware of every cell attached to a tower, where there can be 2, 3,
or 6 cells per tower, which is information they had obtained beforehand.
They found both these common algorithms "performed very poorly" overall:
a. Strongest RSS (fails when the tower isn't close to the roadway)
b. Weighted Centroid (geometric centers, weighted by signal strength)
The median error in the downtown area, for example, was found to be:
a. Strongest RSS = 2.75km median error
b. Weighted Centroid = 2.83km median error
The performance in rural areas was vastly different due likely, they said,
to the fact that the towers are most likely closer to the roadways.
a. Strongest RSS = 7.0km median error
b. Weighted Centroid = 0.7km median error
Their conclusion was that "blindly applying these algorithms to estimate
cell tower positions results in very large errors."
There was more, but that's a quick summary up to the point that they delved
into the reasons for the estimation errors, and how they'd eliminate them.
After they looked at the cause of the errors, they concluded "that one can
_hope_ to locate a cell accurately _only_ if it falls within the area
covered by the wardriving trace", and then they proposed ways to do that.
Bear in mind, while they claim it's the first study of its scale, it was
published in 2010 so it's using ancient technology of its day at that time.
--