Setting up a temporary long-distance wifi for a science experiment

Good Day,

I am trying to do an experiment on Bell's Inequality, which is a part
of Quantum Mechanics. In order to conduct the experiment we need to
have two observation stations each several kilometers away from a
sending station, with a stable internet connection between them.

We will have the sending station on the fourth floor of our building,
and the receiving stations will be in opposite directions from the
building. This is not a permanent set-up: we will assemble and align the
receivers each time we wish to do a test, and then pack it up.

Currently I'm planning on using a D-Link DIR-655 (it has three
removable antennas). I'm going to connect two of the antenna jacks to
directional Yagis ('WiFi Antenna | 14 Element WiFi Yagi Antenna'
(http://www.radiolabs.com/products/an...4eleyagi.php)), and
point the Yagis in different directions towards the receivers. On each
receiver ends I'm going to have another Yagi, which will be plugged into
a Buffalo Technology WLICBG54HP wireless adapter which in turn is
connected to a laptop.

Will this setup work? I'm worried about how the router will function.
In order for a computer to connect does the computer need to be in range
of all three antennas? Because obviously in my setup each computer would
be in range of one only. D-Link does not seem to know the answer
either.

If this setup will not work, could someone please recommend some
alternate hardware? My budget is less than $500 Canadian, but other than
that there are no hard restrictions. We'd like a range as long as
possible, but don't have any real target range.

Thank you,
Peter


------------------------------------------------------------------------
View this thread: http://www.wirelessforums.org/showthread.php?t=38838
http://www.wirelessforums.org

IQC Peter <(E-Mail Removed)> wrote:
> I am trying to do an experiment on Bell's Inequality, which is a part
> of Quantum Mechanics. In order to conduct the experiment we need to
> have two observation stations each several kilometers away from a
> sending station, with a stable internet connection between them.

Do you really mean an /Internet/ connection between the devices? (In
which case why not just use a local Internet connection and run some
sort of VPN across the resulting link. OpenVPN works well for me.) Or
do you mean that you require a direct /network/ connection?

Chris

IQC Peter <(E-Mail Removed)> hath wroth:

>Good Day,

Well, it was until you asked an interesting question. Please don't do
that.

>I am trying to do an experiment on Bell's Inequality, which is a part
>of Quantum Mechanics.

<http://en.wikipedia.org/wiki/Bell's_theorem>

>In order to conduct the experiment we need to
>have two observation stations each several kilometers away from a
>sending station, with a stable internet connection between them.

How "stable"? I presume you mean that you're able to do accurate
differential timing measurements between the two stations. That's not
going to happen with wireless. The transmission and reception delays
are NOT constant or stable and will vary substantially. Attach some
numbers to "stable" and I'll offer some suggestions.

>We will have the sending station on the fourth floor of our building,
>and the receiving stations will be in opposite directions from the
>building. This is not a permanent set-up: we will assemble and align the
>receivers each time we wish to do a test, and then pack it up.

1. Do you have line of sight between the two endpoints?
2. Do you have Fresnel Zone clearance? You didn't specify the number
of kilometers, so I can't calculate it for you. See:
<http://www.terabeam.com/support/calculations/fresnel-zone.php>
3. Is there any local interference on 2.4GHz along the path?

>Currently I'm planning on using a D-Link DIR-655 (it has three
>removable antennas). I'm going to connect two of the antenna jacks to
>directional Yagis ('WiFi Antenna | 14 Element WiFi Yagi Antenna'
>(http://www.radiolabs.com/products/an...14eleyagi.php), and
>point the Yagis in different directions towards the receivers.

Won't work. You will NOT get a MIMO based high speed connection at a
distance of several kilometers. At best, you'll get perhaps about a
12Mbit/sec connection, which will yield about half or 6Mbits/sec in
TCP thruput. You only need one antenna at each end as the MIMO
wireless router will revert to 802.11g when confronted with long time
delays (at long ranges).

Also, Yagi antennas are convenient but just don't have the gain that
you can get with a dish. The typical yagi is 14-16dBi, while the dish
goes up to 24dBi.

When you figure out how far are "several kilometers", plug them into a
path loss calculator to see how big an antenna you need and what
speeds you can expect:
<http://wireless.wikia.com/wiki/Wi-Fi#Link_Calculations>

>On each
>receiver ends I'm going to have another Yagi, which will be plugged into
>a Buffalo Technology WLICBG54HP wireless adapter which in turn is
>connected to a laptop.

Good unit. I like Buffalo.

>Will this setup work?

Probably not for what I'm guessing you are attempting. You can
probably time code your packets from a GPS source, but you did't
mention that. You just wanted a "stable internet" connection,
whatever that means. More clues needed.

>I'm worried about how the router will function.
>In order for a computer to connect does the computer need to be in range
>of all three antennas?

Yes. No line of sight, no communications. Actually, with the Fresnel
Zone problem, you'll need MORE than line of sight at 2.4GHz.

>Because obviously in my setup each computer would
>be in range of one only. D-Link does not seem to know the answer
>either.
>
>If this setup will not work, could someone please recommend some
>alternate hardware? My budget is less than $500 Canadian, but other than
>that there are no hard restrictions. We'd like a range as long as
>possible, but don't have any real target range.

Then find a target range. If you must guess, assign a worst case
distance. Otherwise, calculations are impossible. You also need to
put some numbers on:
1. Expected bandwidth (download speed).
2. Expected reliability (outage hrs per day)
3. Line of sight issues.

I strongly suggest you forget about MIMO and all the acronyms that
claim to be better than 802.11g. The problem is that none of these
improve range. They generally improve speed, not range. Details on
request (I'm late for lunch).


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

To Chris:
As long as the computers can communicate, the experiment will work.
Either a network or an internet connection is fine. However we need
line of sight, and where we're planning on during the experiments (in
a field) there is no local internet connection.

To Jeff:
By stable, I just mean that the connection is constant, and it won't
disconnect regularly (losing the connection makes our program crash).
We have GPS-synched time-stampers for the incident photons, so we
don't have to worry about network delays.

We will have line of sight (otherwise our photons wouldn't make it,
which would kill the experiment). I'd set the minimum distance at 2km
from each receiver station to the sending station, for a total of 4km
between the receiving stations. More distance would be great, but that
should be a good start. There will be the occasional tree branch in
the Fresnel zone, but it will be mostly clear. There is mostly likely
some 2.4GHz background radiation, but we're doing this in a field so
it should be minimal.

As far as the MIMO, I think I must have explained myself poorly. Each
receiver station will have only one antenna. However, the central
sending station will need two directional antennas, one pointed at
each receiving station. I just wasn't sure if this would screw up the
router, since each computer would be connecting via only one antenna,
and the router may require all antennas to be in range of the
computer. I know that a satellite dish would offer higher gain,
however we feel that it would be overly cumbersome since we will have
to move the equipment every time we wish to do the experiment. We
thought that Yagis were a good compromise.

We don't need a high speed connection, we're only sending text. Say 50
kB/s. We will be running the experiment exclusively at night (during
the day, the sunlight overpowers our signal photons). We would run the
experiment for around 4 hours a night, say 3 nights a week. It would
be nice if the connection was solid for that time - a disconnection
crashes the program, and it is a bit of a pain to have to restart it.

Thanks for your help.

On Mon, 18 Feb 2008 10:12:53 -0800 (PST), (E-Mail Removed) wrote in
<7a5967da-a679-41d1-ad59-(E-Mail Removed)>:

>By stable, I just mean that the connection is constant, and it won't
>disconnect regularly (losing the connection makes our program crash).
>We have GPS-synched time-stampers for the incident photons, so we
>don't have to worry about network delays.

You'll always have some disruption percentage. To keep it low, you'll
need decent signal-to-noise. See calculations in the main wiki below.

>As far as the MIMO, I think I must have explained myself poorly. Each
>receiver station will have only one antenna. However, the central
>sending station will need two directional antennas, one pointed at
>each receiving station. I just wasn't sure if this would screw up the
>router, since each computer would be connecting via only one antenna,
>and the router may require all antennas to be in range of the
>computer.

That won't work. Standard wireless units only use one antenna at a
time. You'll need two, each with its own antenna, one for each remote
location.

>I know that a satellite dish would offer higher gain,
>however we feel that it would be overly cumbersome since we will have
>to move the equipment every time we wish to do the experiment. We
>thought that Yagis were a good compromise.

Not a "satellite" dish -- high-gain dish antennas aren't much more
cumbersome than yagis, with higher gain and easier to work with.

>We don't need a high speed connection, we're only sending text. Say 50
>kB/s. We will be running the experiment exclusively at night (during
>the day, the sunlight overpowers our signal photons). We would run the
>experiment for around 4 hours a night, say 3 nights a week. It would
>be nice if the connection was solid for that time - a disconnection
>crashes the program, and it is a bit of a pain to have to restart it.

Can't fix the program to make it more robust?

--
Best regards, FAQ for Wireless Internet: <http://Wireless.wikia.com>
John Navas FAQ for Wi-Fi: <http://wireless.wikia.com/wiki/Wi-Fi>
Wi-Fi How To: <http://wireless.wikia.com/wiki/Wi-Fi_HowTo>
Fixes to Wi-Fi Problems: <http://wireless.wikia.com/wiki/Wi-Fi_Fixes>

(E-Mail Removed) wrote:
<snip>

> We have GPS-synched time-stampers for the incident photons, so we
> don't have to worry about network delays.

<snip>

Are you running the Quarknet cosmic ray experiments? If so,
where did you get the PMT paddles?

Michael

I'll look into the dishes then, and be sure to get two routers/wifi
access points. I'll be sure to look into the program as well. And no,
I'm not doing any Quarknet cosmic ray experiments at this time.
Thanks for the assistance.

(E-Mail Removed) hath wroth:

>By stable, I just mean that the connection is constant, and it won't
>disconnect regularly (losing the connection makes our program crash).

That's not going to happen with any form of wireless. For example, if
you have a 30dB fade margin in the wireless links:
<http://wireless.wikia.com/wiki/Wi-Fi#Link_Calculations>
That translates to a 99.9% reliability, which is about 9 hours of
downtime per year. For your 4 hour test period, that's 2.5 minutes of
downtime. Wireless is just not a guaranteed delivery mechanism.
That's why start/stop/resume protocols were invented for downloads on
dialup connections, where the connection might be lost due to someone
picking up the phone, and where starting over makes little sense. Work
on the protocol problem.

>We have GPS-synched time-stampers for the incident photons, so we
>don't have to worry about network delays.

Good because packet loss will wreck that.

>We will have line of sight (otherwise our photons wouldn't make it,
>which would kill the experiment).

Do the Fresnel Zone calculations:
<http://www.terabeam.com/support/calculations/fresnel-zone.php>
The usual goof is having the antennas too low, where the ground enters
the Fresnel Zone and mangles reliability. At 2km range, you'll need 8
meters clearance at midpoint around the line of sight. That's a nice
way of saying that your link antenna must be 8 meters off the ground
(assuming a flat earth model). Are they?

>I'd set the minimum distance at 2km
>from each receiver station to the sending station, for a total of 4km
>between the receiving stations. More distance would be great, but that
>should be a good start. There will be the occasional tree branch in
>the Fresnel zone, but it will be mostly clear.

Fine, but realize that the link calculations in the aformentioned URL
are the ideal and best case. Everything you throw at your link will
result in a reduction in signal strenth, SNR, and/or reliability. It
only gets worse, not better.

>There is mostly likely
>some 2.4GHz background radiation, but we're doing this in a field so
>it should be minimal.

Ahem. See list of possible sources of intereference at:
<http://wireless.wikia.com/wiki/Wi-Fi#Interference>
The plague of municipal wireless networks and home systems are the
major sources of interference. Over a pair of 2km links, presumably
in a metropolitan area, the chances of encountering such interference
is VERY high. It will come and go, so again, your delivery protocol
is important.

>As far as the MIMO, I think I must have explained myself poorly. Each
>receiver station will have only one antenna. However, the central
>sending station will need two directional antennas, one pointed at
>each receiving station.

Can I pass on explaining how MIMO (802.11n Draft 2) works and simply
state that MIMO is about speed, not range. It won't guy you anything
except at close range. Details later if you want them.

>I just wasn't sure if this would screw up the
>router, since each computer would be connecting via only one antenna,
>and the router may require all antennas to be in range of the
>computer. I know that a satellite dish would offer higher gain,
>however we feel that it would be overly cumbersome since we will have
>to move the equipment every time we wish to do the experiment. We
>thought that Yagis were a good compromise.

It doesn't matter. Yagi, dish, panel, coffee can, or paper clip
antennas are not going to make MIMO work at 2km range. The crude
explanation is that range and speed are inversely proportional (at a
given power level and bandwidth). See range chart at:
<http://wireless.wikia.com/wiki/Wi-Fi#Performance_and_Speed>
Roughly, if you double the distance, you cut the speed to 1/4th. That
includes MIMO (with a few exceptions that hog more bandwidth). By the
time you get past about 25 meters range, all the "enhanced" modes
beyond 54Mbits/sec crap out. Forget about MIMO.

I explained why a yagi sucks as compared to a dish in my previous
rant. Please re-read.

>We don't need a high speed connection, we're only sending text. Say 50
>kB/s.

Then you might be using the wrong hardware and technology. Wi-Fi is
all about speed. At low speeds, there are 900MHz point to point
radios, that will work much better. Some vendors:
<http://www.freewave.com>
<http://www.avalanwireless.com/product_list.htm>
<http://www.maxstream.net>
etc. Search for "900Mhz RS-232" or "900MHz RF modem".

These are more expensive but will work much better. At 900Mhz, the
yagi is the antenna of choice. I'm not sure what you're using for
interface, but they come in ethernet or serial i/o. They are also
commonly used by surveyors for Diff GPS correction links. I have a
few of the Freewave radios sending telemetry (weather, etc) over an
8km link with lousy line of sight.

If you know anyone around that does wireless data collection or
telemetry, they probably have at least one pair of link radios that
can be borrowed.

>We will be running the experiment exclusively at night (during
>the day, the sunlight overpowers our signal photons).

Yep. Science is best done under cover of darkness.

>We would run the
>experiment for around 4 hours a night, say 3 nights a week. It would
>be nice if the connection was solid for that time - a disconnection
>crashes the program, and it is a bit of a pain to have to restart it.

Fix the program. There's no reason for that to happen and there's no
way you're going to get a disconnect free, or zero packet loss
wireless link without protocol assistance.

>Thanks for your help.

Best of luck. Sounds like an interesting experiment.

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