Fastest Bridging Configuration

[milkspill@gmail.com]

Maybe someone can direct me to a previous post, but simply, I'm
looking for the fastest wireless link I can set up (bridging). I have
two LANs (separate buildings) that are 0.4 miles apart that I want to
bridge together.

I've set up past links using Buffalo WHR-HP-G54s running Tomato
Firmware. I get low latency with 25Mbps max of throughput (with semi-
parabolic grid antennas - 24dBi). Given that Buffalo can't sell
wireless equipment due to an injunction, I cannot go that route
again. Besides, I'd like to go faster.

I've thought up some ideas on ways to leverage standard 802.11
equipment. Some thoughts:

1. Channel bond A & G radios in a dual-radio bridge, such as the
Cisco Aironet 1240 (no success there yet).

2. Run two bridge links and bond the two links at each end with a
capable switch (Cisco calls it "EtherChannel).

3. Bridge with 802.11n (Draft 2.0). I understand that some of the
new technologies of N still are effective in narrow focused beam
configuration (24 dBi SPG antennas) over G. Maybe use antennas that
have broader coverage if MIMO can be taken advantaged at that
distance.

4. Full-duplex (Not to spec, I know) / transmit on 1 antenna, receive
on another (with difference polarization). Or use multiple radios,
one that transmit on 5 ghz, but receives on 2.4 ghz (is that even
possible?)

I'd like to hear of some (better) ideas, especially what has actually
worked. I can probably spend a $1000 on each end for the radios if I
can get better speeds.

Thanks,

Michael

[Aaron Leonard]

You could try getting a pair of AIR-AP1252AG-A-K9 (under a grand apiece
street price.) 24dBi is overkill for this app; Yagis make sense. (3 per
AP, ideally a couple feet apart. Should be able to get 100Mbps HDX at least?
(Remember to use AES if you want 11n rates.)

Should ought to work, although you'd be something of a pioneer.

Cheers,

Aaron

----

~ Maybe someone can direct me to a previous post, but simply, I'm
~ looking for the fastest wireless link I can set up (bridging). I have
~ two LANs (separate buildings) that are 0.4 miles apart that I want to
~ bridge together.
~
~ I've set up past links using Buffalo WHR-HP-G54s running Tomato
~ Firmware. I get low latency with 25Mbps max of throughput (with semi-
~ parabolic grid antennas - 24dBi). Given that Buffalo can't sell
~ wireless equipment due to an injunction, I cannot go that route
~ again. Besides, I'd like to go faster.
~
~ I've thought up some ideas on ways to leverage standard 802.11
~ equipment. Some thoughts:
~
~ 1. Channel bond A & G radios in a dual-radio bridge, such as the
~ Cisco Aironet 1240 (no success there yet).
~
~ 2. Run two bridge links and bond the two links at each end with a
~ capable switch (Cisco calls it "EtherChannel).
~
~ 3. Bridge with 802.11n (Draft 2.0). I understand that some of the
~ new technologies of N still are effective in narrow focused beam
~ configuration (24 dBi SPG antennas) over G. Maybe use antennas that
~ have broader coverage if MIMO can be taken advantaged at that
~ distance.
~
~ 4. Full-duplex (Not to spec, I know) / transmit on 1 antenna, receive
~ on another (with difference polarization). Or use multiple radios,
~ one that transmit on 5 ghz, but receives on 2.4 ghz (is that even
~ possible?)
~
~ I'd like to hear of some (better) ideas, especially what has actually
~ worked. I can probably spend a $1000 on each end for the radios if I
~ can get better speeds.
~
~ Thanks,
~
~ Michael

[Jeff Liebermann]

On Fri, 16 May 2008 12:26:53 -0700 (PDT), milkspill@gmail.com wrote:

>Maybe someone can direct me to a previous post, but simply, I'm
>looking for the fastest wireless link I can set up (bridging). I have
>two LANs (separate buildings) that are 0.4 miles apart that I want to
>bridge together.

That's about 2100 ft.

>I've set up past links using Buffalo WHR-HP-G54s running Tomato
>Firmware. I get low latency with 25Mbps max of throughput (with semi-
>parabolic grid antennas - 24dBi). Given that Buffalo can't sell
>wireless equipment due to an injunction, I cannot go that route
>again. Besides, I'd like to go faster.

You can squeeze out a bit more using the various 108 or 125Mbit/sec
consumer routers. For example the Buffalo WHR-G125.

>1. Channel bond A & G radios in a dual-radio bridge, such as the
>Cisco Aironet 1240 (no success there yet).

I haven't seen anything like that in a single box. It can be done
with two boxes, but not easily.

>2. Run two bridge links and bond the two links at each end with a
>capable switch (Cisco calls it "EtherChannel).

<http://en.wikipedia.org/wiki/EtherChannel>
<http://en.wikipedia.org/wiki/Link_aggregation>
Look for 802.3ad or Link Aggregation Control Protocol

I've used two parallel 2.4GHz bridges for redunancy and reliability.
As an added bonus, the total thruput was considerably higher than with
a single link. I tried something cute that didn't work. I initially
had an ordinary ethernet switch at each end. I just plugged two
wireless links into both ends expecting thruput to be roughly twice
that of a single link. It was faster, but not quite twice. The
ethernet switches did not provide the necessary load balancing. SNMP
showed that most of the traffic was going via one or the other
wireless link, while the other port just sat there.

>3. Bridge with 802.11n (Draft 2.0). I understand that some of the
>new technologies of N still are effective in narrow focused beam
>configuration (24 dBi SPG antennas) over G. Maybe use antennas that
>have broader coverage if MIMO can be taken advantaged at that
>distance.

I've been thinking of going into the antenna business. I hope you
like that way 3 antennas look at each end. 24dBi is overkill for
2000ft range. Methinks about 12dBi would be quite sufficient.

>4. Full-duplex (Not to spec, I know) / transmit on 1 antenna, receive
>on another (with difference polarization). Or use multiple radios,
>one that transmit on 5 ghz, but receives on 2.4 ghz (is that even
>possible?)

Yes, it's possible with non-802.11 implimentations. For example,
Proxim Lynx radios split the 5.7GHz band in half. One half goes in
one direction, while the other band goes in the other. Real full
duplex (with a cavity duplexer on the back of the Lynx radios).

The way it works with 802.11 is that you can simulate a full duplex
connection by going twice as fast on the wireless link. The radios
switch direction depending on which packets are going in which
direction. With buffering at each end, it looks just like full duplex
at half the wireless rate.

>I'd like to hear of some (better) ideas, especially what has actually
>worked. I can probably spend a $1000 on each end for the radios if I
>can get better speeds.

I don't think you can do it right for $1000/end. Maybe with used
hardware. Look for 5.7GHz or higher frequency hardware. For example:
<http://www.winncom.com/products/category/WPP/list.html>
Orthogon (Motorola) will go at 44Mbits/sec ethernet thruput up to
about 3 miles:
<http://www.motorola.com/ptp>
<http://www.winncom.com/pdf/DS_OS-Gemini_OS-Gemini_Lite_US.pdf>
If you go to 22Ghz or 60GHz, the available speeds are even faster.
125Mbits/sec ethernet thruput is possible:
<http://www.winncom.com/products/manuf/10110/list.html>
Just, don't ask the prices.

After you've done the requisite window shopping, look on eBay for
various Buffalo wireless products. Search for "buffalo wireless".
There are plenty available (at rediculously high prices).
--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558 jeffl@comix.santa-cruz.ca.us
# http://802.11junk.com jeffl@cruzio.com
# http://www.LearnByDestroying.com AE6KS

[DTC]

milkspill@gmail.com wrote:
> 4. Full-duplex (Not to spec, I know) / transmit on 1 antenna, receive
> on another (with difference polarization). Or use multiple radios,
> one that transmit on 5 ghz, but receives on 2.4 ghz (is that even
> possible?)

We use two 5.8 Gig units at each end with the 22 dBi BBQ grid antennas
on the transmitters to keep them legal and 32 dBi solid dish antennas on
the receivers.

You can use 5.8 Gig, but there are only two non-overlapping channel
configurations that you could use: 149 & 162, or 153 & 165, or
possibly three: 149, 157, 165 if you deploy it right. Or you
could use 5.8/2.4 combo.

If reliability is not an issue, you could use any one of lower power
48 - 64 channels at one end, and any one of the higher power 149 - 165
at the other end.

Even with a grid or dish antenna, you need to keep them at least or
about three feet apart to prevent desensing of the receivers.

[milkspill@gmail.com]

I appreciate all of the replies and various Ideas.

On May 16, 4:16*pm, Jeff Liebermann <je...@cruzio.com> wrote:
> On Fri, 16 May 2008 12:26:53 -0700 (PDT), milksp...@gmail.com wrote:
> >Maybe someone can direct me to a previous post, but simply, I'm
> >looking for the fastestwirelesslink I can set up (bridging). *I have
> >two LANs (separate buildings) that are 0.4 miles apart that I want to
> >bridge together.
>
> That's about 2100 ft.
>
> >I've set up past links using Buffalo WHR-HP-G54s running Tomato
> >Firmware. *I get low latency with 25Mbps max of throughput (with semi-
> >parabolic grid antennas - 24dBi). *Given that Buffalo can't sell
> >wirelessequipment due to an injunction, I cannot go that route
> >again. *Besides, I'd like to go faster.
>
> You can squeeze out a bit more using the various 108 or 125Mbit/sec
> consumer routers. *For example the Buffalo WHR-G125.
>
> >1. *Channel bond A & G radios in a dual-radio bridge, such as the
> >Cisco Aironet 1240 (no success there yet).
>
> I haven't seen anything like that in a single box. *It can be done
> with two boxes, but not easily.
>
> >2. *Run two bridge links and bond the two links at each end with a
> >capable switch (Cisco calls it "EtherChannel).
>
> <http://en.wikipedia.org/wiki/EtherChannel>
> <http://en.wikipedia.org/wiki/Link_aggregation>
> Look for 802.3ad or Link Aggregation Control Protocol
>
> I've used two parallel 2.4GHz bridges for redunancy and reliability.
> As an added bonus, the total thruput was considerably higher than with
> a single link. *I tried something cute that didn't work. *I initially
> had an ordinary ethernet switch at each end. *I just plugged twowirelesslinks into both ends expecting thruput to be roughly twice
> that of a single link. *It was faster, but not quite twice. *The
> ethernet switches did not provide the necessary load balancing. *SNMP
> showed that most of the traffic was going via one or the otherwirelesslink, while the other port just sat there.

Was this with LACP?

>
> >3. *Bridge with 802.11n (Draft 2.0). *I understand that some of the
> >new technologies of N still are effective in narrow focused beam
> >configuration (24 dBi SPG antennas) over G. Maybe use antennas that
> >have broader coverage if MIMO can be taken advantaged at that
> >distance.
>
> I've been thinking of going into the antenna business. *I hope you
> like that way 3 antennas look at each end. *24dBi is overkill for
> 2000ft range. *Methinks about 12dBi would be quite sufficient.
>
> >4. *Full-duplex (Not to spec, I know) / transmit on 1 antenna, receive
> >on another (with difference polarization). *Or use multiple radios,
> >one that transmit on 5 ghz, but receives on 2.4 ghz (is that even
> >possible?)
>
> Yes, it's possible with non-802.11 implimentations. *For example,
> Proxim Lynx radios split the 5.7GHz band in half. *One half goes in
> one direction, while the other band goes in the other. *Real full
> duplex (with a cavity duplexer on the back of the Lynx radios).
>
> The way it works with 802.11 is that you can simulate a full duplex
> connection by going twice as fast on thewirelesslink. *The radios
> switch direction depending on which packets are going in which
> direction. *With buffering at each end, it looks just like full duplex
> at half thewirelessrate.
>
> >I'd like to hear of some (better) ideas, especially what has actually
> >worked. *I can probably spend a $1000 on each end for the radios if I
> >can get better speeds.
>
> I don't think you can do it right for $1000/end. *Maybe with used
> hardware. *Look for 5.7GHz or higher frequency hardware. *For example:
> <http://www.winncom.com/products/category/WPP/list.html>
> Orthogon (Motorola) will go at 44Mbits/sec ethernet thruput up to
> about 3 miles:
> <http://www.motorola.com/ptp>
> <http://www.winncom.com/pdf/DS_OS-Gemini_OS-Gemini_Lite_US.pdf>
> If you go to 22Ghz or 60GHz, the available speeds are even faster.
> 125Mbits/sec ethernet thruput is possible:
> <http://www.winncom.com/products/manuf/10110/list.html>
> Just, don't ask the prices.
>

I'm looking at Trango's TrangoLINK-45 -- they claim 45Mbps of
throughput for a set of radio for $2000 (promo right now) w/integrated
23dBi antenna.
http://www.trangobroadband.com/wireless_products/TrangoLINK_45_usa.shtml


> After you've done the requisite window shopping, look on eBay for
> various Buffalowirelessproducts. *Search for "buffalowireless".
> There are plenty available (at rediculously high prices).
> --
> # Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
> # 831-336-2558 * * * * * *je...@comix.santa-cruz.ca.us
> #http://802.11junk.com* * * * * * * je...@cruzio.com
> #http://www.LearnByDestroying.com* * * * * * * AE6KS

[Jeff Liebermann]

On Mon, 19 May 2008 12:06:08 -0700 (PDT), milkspill@gmail.com wrote:

>Was this with LACP?

No. It was with two junk ethernet switches that did not have 802.3ad
features:
<http://en.wikipedia.org/wiki/Link_Aggregation_Control_Protocol>
It might have worked if I had fancier switches available.

>I'm looking at Trango's TrangoLINK-45 -- they claim 45Mbps of
>throughput for a set of radio for $2000 (promo right now) w/integrated
>23dBi antenna.
>http://www.trangobroadband.com/wireless_products/TrangoLINK_45_usa.shtml

I've use other Trango products. Crude documentation, limited
monitoring (SNMP), marginal support, some quality control problems. I
wasn't thrilled, but it eventually did work for me.

Well, you could get 50Mbits/sec using two "aggregated?" 802.11g links
at 25Mbits/sec each. The cost would be much less than $2,000, but
would be rather messy with two antennas at each end.

--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558 jeffl@comix.santa-cruz.ca.us
# http://802.11junk.com jeffl@cruzio.com
# http://www.LearnByDestroying.com AE6KS

[Bod43@hotmail.co.uk]

On 20 May, 04:36, Jeff Liebermann <je...@cruzio.com> wrote:
> On Mon, 19 May 2008 12:06:08 -0700 (PDT), milksp...@gmail.com wrote:
> >Was this with LACP?
>
> No. *It was with two junk ethernet switches that did not have 802.3ad
> features:
> <http://en.wikipedia.org/wiki/Link_Aggregation_Control_Protocol>
> It might have worked if I had fancier switches available.
>
> >I'm looking at Trango's TrangoLINK-45 -- they claim 45Mbps of
> >throughput for a set of radio for $2000 (promo right now) w/integrated
> >23dBi antenna.
> >http://www.trangobroadband.com/wireless_products/TrangoLINK_45_usa.shtml
>
> I've use other Trango products. *Crude documentation, limited
> monitoring (SNMP), marginal support, some quality control problems. *I
> wasn't thrilled, but it eventually did work for me.
>
> Well, you could get 50Mbits/sec using two "aggregated?" 802.11g links
> at 25Mbits/sec each. *The cost would be much less than $2,000, but
> would be rather messy with two antennas at each end.
>
> --
> # Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
> # 831-336-2558 * * * * * *je...@comix.santa-cruz.ca.us
> #http://802.11junk.com* * * * * * * je...@cruzio.com
> #http://www.LearnByDestroying.com* * * * * * * AE6KS

Here are some nice 1G laser units that I looked up for fun :-

Geodesy FSO Pronto SuperGIG 1000Base-T copper RJ45 Interface
Maximum 1000Mtrs.
ONLY £14390.00 + VAT Pair

A bit over budget?

[Jeff Liebermann]

On Tue, 20 May 2008 12:06:50 -0700 (PDT), Bod43@hotmail.co.uk wrote:

>Here are some nice 1G laser units that I looked up for fun :-
>
>Geodesy FSO Pronto SuperGIG 1000Base-T copper RJ45 Interface
>Maximum 1000Mtrs.
>ONLY £14390.00 + VAT Pair
>
>A bit over budget?

Budgets are made to be broken. I can proudly say that I've never even
come close to staying within my original wild guess proposal estimate.
I don't mind being wrong, as long as I'm consistently wrong.

Thanks for reminding me about FSO (free space optics). There are
slower and cheaper LED based solutions. I've used FSO boxes from
Plaintree in Canada:
<http://www.plaintree.com/products/>
<http://www.plaintree.com/faqs.html>
in the past with good results. The big problems are fog, direct
sunlight pollution, stupid birds, and parnoid neighbors. At 2000ft,
even their lowest end "indoor" boxes will work. The only prices I
could find was from 2000 when a customer bought a pair of something
for about $9,000/pair. I guess it's lower now, but I can't find any
current pricing.

I helped setup a pair of these:
<http://www.lightpointe.com/products/fl_100.cfm>
(in the rain). They work nicely when the lenses aren't wet. Here's
one (or two?) on eBay:
<http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=260242185588>
for $2,000.

More:
<http://en.wikipedia.org/wiki/Free_space_optics>

--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558 jeffl@comix.santa-cruz.ca.us
# http://802.11junk.com jeffl@cruzio.com
# http://www.LearnByDestroying.com AE6KS