An
important point to clarify is that the precision is not the application’s, but
the device’s. In other words, the GPS device of the iPhone or iPad.
For that
reason I tried to search the internet, and also on Apple's website, for
information on its accuracy. But my search was in vain. I did not find any
relevant information about GPS accuracy. So I decided to use what I had in hands
to come to some conclusion.
What I did
was to create an application that collected the GPS’s coordinates, as well as
the accuracy of each coordinate obtained, for which I could note the conditions
of the surrounding places where they were collected, and save them for later
analysis. I developed this App in a very simple and quick way to make this kind
of surveys, and then I left to the street...
I will
start backwards, first presenting the summary and conclusion of the survey.
Methodology Summary
I’ve collected
a little more than 2000 points during one week of walking and car rides through
several different places of the city, with different surroundings. The
survey was conducted in the city of Vitória/ES, Brazil, in April 2017. I have
chosen open locations such as the sidewalk along the beach, where the nearest
buildings are around 100 meters away. But I also walked in the middle of the city, on
the sidewalk or in the middle of the street, but with buildings on both sides.
And I also collect points at the top of one of the city's postcards, the
Convent of Penha, which sits atop a hill about 500 meters above sea level, with
nothing else around.
As the
messages we see on the iPhone itself suggest that the accuracy is better when
Wi-Fi is turned on, I understood that I should collect coordinates with the
Wi-Fi switched on and off, and also with the 3G/4G switched
on and off, and also with the cell phone plugged into the charger or not, that
is, in several different ways so we could come up with some more effective
conclusion about accuracy.
Conclusions
Starting
backwards...
The best
horizontal accuracy achieved was 5 meters.
The best
vertical accuracy achieved was 3 meters.
Measurements
were always between 98% and 99% of the horizontal accuracy maximum, and around
92% of the vertical accuracy maximum, in regions where there was a high
visibility of the sky, what we use to call as open sky. What means that there
were no barriers on satellite’s visibility like buildings around the places where
the measurements were taken.
On the
other hand the accuracy varied greatly when the coordinates were taken between
buildings, where satellite’s visibility was restricted, ranging from 32% to 90%,
depending even if the Wi-Fi and 3G/4G were switched on or off. On this point
it is worth to see the details of the analysis presented below.
Segmenting the Analysis
I took into
consideration that the optimum horizontal precision is 5 meters and the optimum
vertical precision is less than or equal to 6 meters. The analysis presented
from this point on will follow these values.
I have
segmented the measurements collection results into:
·
Wi-Fi
Switched On or Off;
·
3G/4G
Switched Or r Off;
·
Device
charging or not charging;
And we came
to the following conclusions:
Wi-Fi Switched On or Off?
Analyzing
data considering only whether Wi-Fi was on or off:
·
Horizontal Accuracy:
o
In
open sky locations having the Wi-Fi switched on or off did not make any difference in the
horizontal accuracy of the collected points. A total of 98% of the points were
collected at maximum precision;
o
In
places surrounded by buildings, having the Wi-Fi off was a bit more accurate than the
other way around. But the difference was small: 83% of maximum accuracy with
Wi-Fi off versus 80% with Wi-Fi on. Almost a technical
tie;
·
Vertical Accuracy:
o
In
places with open skies the behavior of the vertical precision was similar to
the horizontal precision. It had no difference if the Wi-Fi was switched on or
off, when we've got 92% of the points collected at the highest precision;
o
In
places surrounded by buildings, having the Wi-Fi switched on brought better
accuracy. We had 73% maximum accuracy with Wi-Fi on versus 63% with Wi-Fi off;
3G/4G Switched On or Off?
Analyzing
data considering only whether 3G/4G was on or off:
·
Horizontal Accuracy
o
Once
more, in places with the sky open, having the 3G/4G turned on or off, whether
or not Wi-Fi was on or off, made no difference in the accuracy of the collected
points. Between 98% and 99% of the points were collected at maximum precision;
o
in
places surrounded buildings we had better accuracy with the 3G/4G switched on,
with 85% of points collected with the maximum accuracy, versus 61% when it
was off;
·
Vertical Accuracy
o
In
open sky locations, vertical accuracy was the same with 3G/4G on or off, ranging
from 92% to 93%;
o
In
areas surrounded by buildings the difference was quite large, with 75% of
maximum accuracy with 3G/4G on, compared to 41% with 3G/4G off;
Wi-Fi and 3G/4G Together
We had the following when analyzing
the survey accuracies together with the combinations of Wi-Fi and 3G/4G turned
on or off:
·
Horizontal Accuracy
o
Following
the behavior identified individually, combining both indicators, when the
points were collected in a region of open sky, the accuracy is always very
good, ranging between 98% and 99%;
o
Between
buildings we had better accuracy when the 3G/4G was on and the Wi-Fi off,
getting 90% of the best accuracies. On the other hand the worst measurements
were made with the 3G/4G off and Wi-Fi on. The table below shows all the combined
options:
|
Between Buildings
|
3G
|
||
|
On
|
Off
|
||
|
Wi-Fi
|
On
|
83%
|
54%
|
|
Off
|
90%
|
66%
|
|
·
Vertical Accuracy
o
Under an open sky, the vertical accuracy follows the pattern, regardless of whether
the indicators are on or off, maximum accuracy occurs between 98% and 99% of
the measurements taken;
o
And
again, it varies a lot among buildings: with the Wi-Fi and 3G/4G both switched
on, getting 74% of the best, and the worst, with both switched off, reaching
only 32%. The table below illustrates the options:
|
Between Buildings
|
3G
|
||
|
On
|
Off
|
||
|
Wi-Fi
|
On
|
74%
|
58%
|
|
Off
|
75%
|
32%
|
|
Device charging or not?
Having the
iPhone charging did not make any difference in the accuracy of the points
collected. All results were similar in both conditions.
