On Tue, 2 Aug 2016 09:03:08 +0200, Michael Eyd wrote:

The primary directions for mobile network antennas and WiFi antennas may
be different, so one would have to test them independently...

True. However, unless you use an RF anechoic chamber, the influences
of the room environment will have a bigger influence than the antenna
patterns. Reflectors and absorbers will ruin any test, unless you're
interested in performing a "real world" type of test, which is what
this range test might be considered. For example:
http://www.smallnetbuilder.com
does their benchmarks indoors, with plenty of walls and furniture to
get in the way. I think it's part of Tim Higgins house, but I'm not
su
http://www.smallnetbuilder.com/wireless/wireless-features/32512-does-an-ac-router-improve-n-device-performance
The RF environment is far from perfect, but it's identical for each
router being tested, which the point of the test:

More on how they run their tests:
http://www.smallnetbuilder.com/wireless/wireless-howto/32478-how-we-test-wireless-products-revison-8
http://www.smallnetbuilder.com/wireless/wireless-howto/32993-how-we-test-wireless-products-revision-9
http://www.smallnetbuilder.com/wireless/wireless-howto/32944-how-we-test-mu-mimo
and even mo
http://www.smallnetbuilder.com/tags/how-we-test

The overall results are rather interesting (to me). Different
routers, which use the same chipset and roughly the same antennas,
produce substantially different performance results. I don't have
time to speculate on why, but let's just say that there are is a large
list of uncontrolled factors that have an effect on the measurements.
One can eliminate a fair number with a $100 million RF anechoic
chamber, but that's a bit beyond my present means.

Side-by-side (taking this literally) might be yet another influencing
factor, where one device (might) severely interfere with the other.

True. However, if a wireless client is associated with an access
point, but not passing any traffic other than the usual beacons and
broadcasts, there is very little traffic that might constitute
interference. Offhand, my guess(tm) is about a 1/100 duty cycle. Were
any of these packets to collide with traffic from a nearby wireless
device, the error would be about 1% from the collision.

However, for the range test, this will have no effect because we're
not trying to squeeze as many packets as possible through a pipe.
We're trying to determine the range at which it is no longer possible
to pass packets or where the connection becomes unstable. At worst,
packet collisions will "blurr" the results somewhat. I don't consider
proximity to be a problem.

Additionally (forgot to mention that in my previous post) there
shouldn't be anybody running around inside the test area (which is
larger than just the direct line of sight between the device(s) and the
AP), no cars should be passing in the vicinity, there should be no
neighboring WiFi networks even at the horizon, ...

Part of the range test is take the tablet or iphone and walk away from
the wireless router, noting the range at which traffic ceases.
Presumably, one would need to hold the device to do that. At the
frequencies involved, placing the device on top of a cardboard box
when carrying it will minimize proximity effects and antenna detuning,
while still allowing one monitor the device. It's far from perfect,
but methinks good enough.

I won't do the tests, for several reasons:

- I don't have any Android device available, least several different ones.

Borrow one or invite your friends to the test. Or, are all your
friends Apple users? What a horrible thought.

- Where I live I can easily and at any time find several other WiFi
networks.

Not a problem. You're not trying to maximize throughput, just
determine how far you can operate before it quits. You can do that
with ping, which hogs very little bandwidth, and will not interfere
with the neighbors streaming wireless connection.

- I wouldn't have enough open range (without reflections from other
houses, passings cars, heck there are even electrified railroad tracks
at about 500m distance).

I think you'll find that at 802.11g speeds, with the wireless fixed at
54Mbits/sec, you'll get about 30 meters range. The transition between
working and dead will be quite abrupt, usually within a meter or two.
If you find a straight line path that's about 50 meters long, you
should be ok.

Way too bad conditions for performing such a test.

It doesn't matter. We're comparing two devices, not producing an
absolute measurement. Absolute measurements would be nice, so we
could compare your results with mine and others, but that's not going
to happen without an extremely well controlled environment. However,
when comparing two devices, the conditions are identical, and
therefore the comparison is quite valid.

You probably spent more time finding excuses to not run the test than
it would take to actually perform it. Thanks for at least thinking
about the problems involved.


--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558