Geolocation: Difference between revisions

From Helpful
Jump to navigation Jump to search
Tag: New redirect
 
Tag: Removed redirect
Line 1: Line 1:
#redirect [[Javascript_notes_-_browser_related,_APIs#Geolocation_API]]
 
Geolocation
 
==Methods and accuracy==
 
The methods this might be backed by include:
* '''GPS receiver''' (mobile mostly - other devices tend not to have one)
:: in cities, expect at most ~5m in a relatively open area, at most 10m with trees and tall buildings around, and basically nothing indoor (may fall back to any of the below)
:: tends to update no faster than once per second
 
* '''WiFi based positioning'''
: {{comment|(based on seeing known Wifi hotspots that were previously detected by devices that also ''did'' have their GPS on at the time) [https://en.wikipedia.org/wiki/Wi-Fi_positioning_system] - which puts these APs in ''roughly'' the right place}}
:: assume you won't often get better than 20m accuracy - and assume you get little to no coverage unless you're in a city near a house
::: though indoor it sometimes may be down to ~5m if your specific AP happens to be submitted
 
* '''cell tower mapping''' (mobile only)
: same "as previously seen when you had GPS on" idea as just described with WiFi, but using cell phone towers
: ...which cover a much larger area, tends to be ''worse'', so assume this often won't get much better than 1km
 
* IP geolocation
:: primarily for devices that are broadband users - fixed, or on wifi - because of how the publicly visible address are given out to them.
::: ...on mobile networks you ''do'' get addresses but they change frequently and relate little to location)
:: assume you
::: usually won't get better than 1km
::: sometimes a bit better
::: sometimes a lot worse
:: both better and worse relate to how addresses are assigned and divided. Which itself is sometimes very predictable, frequently much less so, so you can
::: ''likely'' place someone in the right country
::: ''regularly'' in the right state/province
::: ''sometimes'' within dozens of km (so within a city or two), and
::: ''occasionally'' down to a university campus or such (1km, but only because universities often got a large fixed subnet, and someone will submit that to the lookup database)
:: the funny thing is that even if the location ''is'' more accurate, you usually don't know that it is. Lookups will only sometimes estimate the precision, and only sometimes estimate it well. So maybe it's better than 100m, maybe it's as bad as 100km.
:: defeated by VPN
 
 
 
 
For the API in browsers, see [[Javascript_notes_-_browser_related,_APIs#Geolocation_API]]

Revision as of 17:36, 27 June 2024

Geolocation

Methods and accuracy

The methods this might be backed by include:

  • GPS receiver (mobile mostly - other devices tend not to have one)
in cities, expect at most ~5m in a relatively open area, at most 10m with trees and tall buildings around, and basically nothing indoor (may fall back to any of the below)
tends to update no faster than once per second
  • WiFi based positioning
(based on seeing known Wifi hotspots that were previously detected by devices that also did have their GPS on at the time) [1] - which puts these APs in roughly the right place
assume you won't often get better than 20m accuracy - and assume you get little to no coverage unless you're in a city near a house
though indoor it sometimes may be down to ~5m if your specific AP happens to be submitted
  • cell tower mapping (mobile only)
same "as previously seen when you had GPS on" idea as just described with WiFi, but using cell phone towers
...which cover a much larger area, tends to be worse, so assume this often won't get much better than 1km
  • IP geolocation
primarily for devices that are broadband users - fixed, or on wifi - because of how the publicly visible address are given out to them.
...on mobile networks you do get addresses but they change frequently and relate little to location)
assume you
usually won't get better than 1km
sometimes a bit better
sometimes a lot worse
both better and worse relate to how addresses are assigned and divided. Which itself is sometimes very predictable, frequently much less so, so you can
likely place someone in the right country
regularly in the right state/province
sometimes within dozens of km (so within a city or two), and
occasionally down to a university campus or such (1km, but only because universities often got a large fixed subnet, and someone will submit that to the lookup database)
the funny thing is that even if the location is more accurate, you usually don't know that it is. Lookups will only sometimes estimate the precision, and only sometimes estimate it well. So maybe it's better than 100m, maybe it's as bad as 100km.
defeated by VPN



For the API in browsers, see Javascript_notes_-_browser_related,_APIs#Geolocation_API