Outdoor Multi-point Wireless

I'm looking to connect three farm houses using Buffalo .11g gear
(using WDS) to share data and share a cable internet connection. The
distances of the paths are 1.25 miles each. I have LOS for the paths.
I plan to use relatively short (<15') lengths of LMR-400 for the
outdoor antennas. I want the best signal (who doesn't?) so I can get
the most from the 11g, realizing that WDS will cut my bandwidth in
half. I figured I'd use a router with a directional antenna at the
house with the cable available, an AP with an omni at the middle house
(it would also have clients connecting wirelessly), and either an AP
(for multiple computers) or an ethernet converter (for a single
computer) with a directional antenna at the third house:

cable/router/directional->-------<AP/omni>-
-
about 125 -
degrees -
-<AP/directional

My questions:
What is the proper gain needed for each antenna?
Can the gain be too great?
Do the antenna gains need to be "balanced" some way?
Should I consider a sector antenna for the middle AP?
What grounding, if any, is needed? The antennas will be at or below
the roof lines of the houses.
What length of LMR-400 could I use without losing too much signal?
LOS between the last two houses may be encrouched upon by a rotating
irrigation boom (4" aluminum pipe plus some smaller pipes). Will this
completely destroy the LOS for this segment?

TIA,
R

Calculation Results


Free Space Loss dB Rx Signal Level dBm
Theoretical System Operating Margin dB







"Consultant" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed) m...
> I'm looking to connect three farm houses using Buffalo .11g gear
> (using WDS) to share data and share a cable internet connection. The
> distances of the paths are 1.25 miles each. I have LOS for the paths.
> I plan to use relatively short (<15') lengths of LMR-400 for the
> outdoor antennas. I want the best signal (who doesn't?) so I can get
> the most from the 11g, realizing that WDS will cut my bandwidth in
> half. I figured I'd use a router with a directional antenna at the
> house with the cable available, an AP with an omni at the middle house
> (it would also have clients connecting wirelessly), and either an AP
> (for multiple computers) or an ethernet converter (for a single
> computer) with a directional antenna at the third house:
>
> cable/router/directional->-------<AP/omni>-
> -
> about 125 -
> degrees -
> -<AP/directional
>
> My questions:
> What is the proper gain needed for each antenna?
> Can the gain be too great?
> Do the antenna gains need to be "balanced" some way?
> Should I consider a sector antenna for the middle AP?
> What grounding, if any, is needed? The antennas will be at or below
> the roof lines of the houses.
> What length of LMR-400 could I use without losing too much signal?
> LOS between the last two houses may be encrouched upon by a rotating
> irrigation boom (4" aluminum pipe plus some smaller pipes). Will this
> completely destroy the LOS for this segment?
>
> TIA,
> R

1st house to middle house
Based on your figures at 1.25 miles, 15 foot each LMR 400, 2 Lightening
Arrestors
then a 15dbi directional perhaps a yagi with 30 degree beam and an omni with
9dbi
would give you an SOM (system operating margin) of 19.4db based on a -87dbm
receive sensitivty and a -67.5dbm receive signal. An SOM of at least 10 is
required
for a stable link. The setup would be the same from the 3rd house back to
the middle.
You could proably do a 7dbi omni and still get a good link but its up to
you.
This is with clear RF LOS. If the irrigation system gets in front of the
antenna or in the
fresnal zone then it could cause some signal degradation.

This place has some pretty good prices http://www.fab-corp.com/

Im not an expert at this so I welcome expert comments on my solution




dbm

(E-Mail Removed) (Consultant) wrote:

Hmmm. I charge 15% to consult to consultants.


>My questions:
>What is the proper gain needed for each antenna?

You may find the calculator at this site useful:

http://my.athenet.net/~multiplx/cgi-...eless.main.cgi

These are also helpful:

http://www.ecommwireless.com/calculations.html

>Can the gain be too great?

No.

>Do the antenna gains need to be "balanced" some way?

Do you mean the same dBi? No.

>Should I consider a sector antenna for the middle AP?

Not if you plan on using clients locally. Besides, for the distances you are
dealing with, a good omni (16 dBi or more) with a bit of downtilt is fine. Keep
in mind that there is a "cone of silence" under the antenna when placing it.

>What grounding, if any, is needed? The antennas will be at or below
>the roof lines of the houses.

Doesn't matter where the antennas are, you need a good Nema ground. Put the
grounding block as close to the antenna as possible and run solid core ground
wire to the house panel ground point. Using a separate ground point in the
ground is a good recipe for ground imbalance and component damage.

>What length of LMR-400 could I use without losing too much signal?

See the online tools above.

>LOS between the last two houses may be encrouched upon by a rotating
>irrigation boom (4" aluminum pipe plus some smaller pipes). Will this
>completely destroy the LOS for this segment?

No.

On 21 Nov 2004 20:28:06 -0800, (E-Mail Removed) (Consultant) wrote:

>I'm looking to connect three farm houses using Buffalo .11g gear
>(using WDS) to share data and share a cable internet connection.

You realize that WDS is half duplex and therefore cuts your thruput in
half.

>The distances of the paths are 1.25 miles each. I have LOS for the paths.

At 1.25 miles and minimal antennas, my guess(tm) is that you'll get a
9Mbit/sec association with 802.11g. Thruput should be less than half
at about 4Mbits/sec. Add WDS and it gets cut in half again to
2Mbits/sec.

>I plan to use relatively short (<15') lengths of LMR-400 for the
>outdoor antennas. I want the best signal (who doesn't?) so I can get
>the most from the 11g, realizing that WDS will cut my bandwidth in
>half.

Yep. Your actually thruput will be highly dependent on the connection
speed, which is dependent on signal strength (and quality), which is
primarily dependent on antenna gain.

>I figured I'd use a router with a directional antenna at the
>house with the cable available, an AP with an omni at the middle house
>(it would also have clients connecting wirelessly), and either an AP
>(for multiple computers) or an ethernet converter (for a single
>computer) with a directional antenna at the third house:
>
>cable/router/directional->-------<AP/omni>-
> -
> about 125 -
> degrees -
> -<AP/directional

There's another method that *MIGHT* be worth considering instead of
the omni in the middle. Use a power splitter (lose -4dB) and two
directional antennas pointed at each end of the link. If your antenna
gain is more than the power splitter loss, you have a net improvement.
Of course it costs quite a bit more and tends to be ugly. It's major
benifit is that an omni picks up interference from all directions.
Having a directional antenna removes much of the interference. You
can start with an omni (because it's cheaper) but be prepared to
switch to directional antennas if interference becomes an issue.

Another possibility is to use TWO radios at the midpoint, on different
channels, each with a different directional antenna. Also more
expensive, but it eliminates the problem where WDS store-n-forward
cuts the bandwidth in half.

Note that an omni in the middle limits you to using vertical
polarization. However, two pairs of directional antennas can be any
polarization. In general, most of the interference arrives vertically
polarized, so horizontal polarization is a good idea if possible.

>What is the proper gain needed for each antenna?

Do the math.
http://www.ydi.com/calculation/som.php
What you're looking for is an absolute minimum of 10dB fade margin.
The system will work with 0dB of fade margin, but not be particularly
stable. A passing bird would cause the signal to disappear. At 10dB,
you'll 802.11g, if set to 802.11g only (ignore 802.11b), will probably
be running at the slowest speeds. Any interference will make it
worse. Let's see what 10 times as much (20dB) fade margin yields.

Use these for receiver sensitivity at various connection speeds.
They're from the D-Link DI-624 data sheet and appear to be rather
typical.
* 54Mbps OFDM, 10% PER, -68dBm)
* 48Mbps OFDM, 10% PER, -68dBm)
* 36Mbps OFDM, 10% PER, -75dBm)
* 24Mbps OFDM, 10% PER, -79dBm)
* 18Mbps OFDM, 10% PER, -82dBm)
* 12Mbps OFDM, 10% PER, -84dBm)
* 11Mbps CCK, 8% PER, -82dBm)
* 9Mbps OFDM, 10% PER, -87dBm)
* 6Mbps OFDM, 10% PER, -88dBm)
* 5.5Mbps CCK, 8% PER, -85dBm)
* 2Mbps QPSK, 8% PER, -86dBm)
* 1Mbps BPSK, 8% PER, -89dBm)

Using rather conservative guesswork (better safe than suprised):
Distance = 1.25 miles
TX power = +15dBm (my measurements)
RX sens = -87dBm (at 9Mbits/sec OFDM)
TX ant gain = 15dBi (large panel or small dish)
RX ant gain = 8dBi (cheapo omni)
TX coax loss = -3dB (15ft LMR-400 plus connectors and pigtail)
RX coax loss = -3dB (assumed similar installation)

Fade margin is 12.9dB. That's way too close to be comfortable. You
need to squeeze some decibels out of something. Getting rid of the
omni and using a higher gain antenna is one way. Increasing the size
of the TX antenna from a 15dBi dish to a 24dBi dish is another.
Whatever it takes to get closer to 20dB fade margin.

>Can the gain be too great?

No, but there are side effects. Higher gain antennas have a much
narrower beamwidth. As the gain gets high (over 15dBi), alignment
becomes criticial. In the case of omnis, it's the vertical radiation
angle that gets narrow. If there's an elevation difference, a
rediculously high gain omni may have such a narrow beam width that the
signal goes over the head of the intended target. Watch the beamwidth
and downtilt specs carefully.

>Do the antenna gains need to be "balanced" some way?

No.

>Should I consider a sector antenna for the middle AP?

No. The 125 degree angle is at the limit of sector antennas. Sector
antennas have a different purpose. They're made of "area" coverage
and not for point to point.

>What grounding, if any, is needed? The antennas will be at or below
>the roof lines of the houses.

None. If you live in lightning infested areas, you might consider a
lightning arrestor. The coax cable usually provides all the grounding
that's necessary to prevent static electricity buildup (from the
wind), but even that's minor.

>What length of LMR-400 could I use without losing too much signal?

Do the math (and use the SOM calculator). LMR-400 is about 0.07dB/ft.
I use -1dB loss for each connector pair. Lightning arrestors lose
about 1dB plus two connector pairs. If you want to use more coax,
you'll just need to get a bigger antenna.

>LOS between the last two houses may be encrouched upon by a rotating
>irrigation boom (4" aluminum pipe plus some smaller pipes). Will this
>completely destroy the LOS for this segment?

Probably not. The beamwidth at 1.25 miles is quite wide. Assuming a
24dBi antenna with a -3dB beamwidth of 5 degrees, the signal is about
570 ft wide. There's no way a 4" pipe is going to block all the
signal. There may be some weird multipath effects, but I doubt if
there will be any blockage.


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

Jeff, I agree that 4 directional antennas would be a better option but I
was wondering what your hardware choices would be for this compared to the
wds solution. Obviously they have to be a bridge with ap capabilities or
something like it.

Im not questioning your expertise, just trying to learn more. I am a
wireless airhead student.

And my thought on the 2mbs using wds is that this gives all 3 users about
666mbs or so simultaneous use and if the objective is to share internet
access this is plenty for most ISP speeds at 256 or 512mbs .

If cost were a major factor here wouldnt WDS be the best solution?
..

"Consultant" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed) m...
> I'm looking to connect three farm houses using Buffalo .11g gear
> (using WDS) to share data and share a cable internet connection. The
> distances of the paths are 1.25 miles each. I have LOS for the paths.
> I plan to use relatively short (<15') lengths of LMR-400 for the
> outdoor antennas. I want the best signal (who doesn't?) so I can get
> the most from the 11g, realizing that WDS will cut my bandwidth in
> half. I figured I'd use a router with a directional antenna at the
> house with the cable available, an AP with an omni at the middle house
> (it would also have clients connecting wirelessly), and either an AP
> (for multiple computers) or an ethernet converter (for a single
> computer) with a directional antenna at the third house:
>
> cable/router/directional->-------<AP/omni>-
> -
> about 125 -
> degrees -
> -<AP/directional
>
> My questions:
> What is the proper gain needed for each antenna?
> Can the gain be too great?
> Do the antenna gains need to be "balanced" some way?
> Should I consider a sector antenna for the middle AP?
> What grounding, if any, is needed? The antennas will be at or below
> the roof lines of the houses.
> What length of LMR-400 could I use without losing too much signal?
> LOS between the last two houses may be encrouched upon by a rotating
> irrigation boom (4" aluminum pipe plus some smaller pipes). Will this
> completely destroy the LOS for this segment?
>
> TIA,
> R

On Mon, 22 Nov 2004 15:46:15 -0600, "Airhead"
<(E-Mail Removed)> wrote:

>Jeff, I agree that 4 directional antennas would be a better option but I
>was wondering what your hardware choices would be for this compared to the
>wds solution. Obviously they have to be a bridge with ap capabilities or
>something like it.

No, not so obviously. It can be done with routers. The problem is
that it's complex and more expensive than WDS. A transparent bridge
would NOT be necessary unless the OP wants all 3 locations to appear
on the same LAN for network browsing. That makes lots of sense for a
corporate remote office connection, but not much if you're just
sharing the bandwidth with a few friends. So, let's take it one at a
time.

wAN=IP from ISP WAN=192.168.1.2
LAN=192.168.1.1 LAN=192.168.5.1
DSL==[router]==[Access]==>> <<==[Access]==[router]==[Access]===>>
WAN [ ] [Point ] CH1 [Point ] [ ] [Point ] CH6
[ ] [Client] [ ]
[ LAN ] [Mode ] [ LAN ]
192.168.1.100 192.168.5.100
192.168.1.101 192.168.5.101
192.168.1.102 192.168.5.102
etc etc
Location #1 Location #2

WAN=192.168.5.2
LAN=192.168.8.1
from <<====[Access]==[router]
Location #2 [Point ] [ ]
CH6 [Client] [ ]
[Mode ] [ LAN ]
192.168.8.100
192.168.8.101
192.168.8.102
etc
Location #3

Yes, the router can be combined with the access point to use a single
box. That will work if the coax cable is short or if you wanna put
the router on the roof. I don't mind putting access points on the
roof, but routers involve too many cables.

Only one MAC address need by bridged across the two wireless links, so
you do NOT need to use a transparent bridge. A bridge may be used but
the topology and IP address layout is totally different. Many access
points have a client mode (DWL-900AP+, DWL-2100, WAP54G, WAP11).

In case you're lost:

Location #1
Router
WAN IP = from ISP
LAN IP = 192.168.1.1
Clients
IP = 192.168.1.3 -> 254
GW = 192.168.1.1
NM = 255.255.255.0

Location #2
Router
WAN IP = 192.168.1.2
LAN IP = 192.168.5.1
Clients
IP = 192.168.5.3 -> 254
GW = 192.168.5.1
NM = 255.255.255.0

Location #3
Router
WAN IP = 192.168.5.2
LAN IP = 192.168.8.1
Clients
IP = 192.168.8.3 -> 254
GW = 192.168.8.1
NM = 255.255.255.0

Note that this daisy chain type of system can extend indefinately.
The midpoint (Location #2) has the advantage of being able to transmit
and receive at the same time. Therefore max thruput is maintained.

>Im not questioning your expertise, just trying to learn more. I am a
>wireless airhead student.

Groan...

>And my thought on the 2mbs using wds is that this gives all 3 users about
>666mbs or so simultaneous use and if the objective is to share internet
>access this is plenty for most ISP speeds at 256 or 512mbs .

I'm not sure exactly how WDS works. I've only played with one WDS
system and it was sufficiently complex to make measurements difficult.
My guess(tm), is that connectivity between Location #1 and Location #2
are full speed as limited by S/N ratio, range, etc. Traffic to
Location #3 goes full speed between #1 and #2, but half speed between
#2 and #3 because everything has to transmitted twice with WDS.

It gets a bit messy when both #2 and #3 are downloading. Fair share
says that traffic will be equally split at #1. Therefore, traffic to
#2 will get 50% of the bandwidth. The rest gets cut in half by #2
WDS, so that #3 gets only 1/4th the bandwidth. I'm not sure about
this, but it seems like a good first guess.

>If cost were a major factor here wouldnt WDS be the best solution?

Since when has cost NOT been a factor? Good, Fast, Cheap...pick two.
Yes, WDS is the cheapest. I don't think you want to hear my rant on
wasted traffic, polluting the airwaves with duplicate transmissions,
and efficiency (that also applies to mesh networks).


--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831.336.2558 voice http://www.LearnByDestroying.com
# (E-Mail Removed)
# 831.421.6491 digital_pager (E-Mail Removed) AE6KS