Replacing DSL with Pt-to-Pt wireless link to FiOS

After significant troubleshooting I conclude that the 'migration' that
Megapath conducted in early 2011 to former Speakeasy DSL backhauls has
permanently limited my line speed to 40% of the rated speed and former
actual speed.

As a result, I would like to replace the static IP DSL circuit with a
WiFi-b/g pt-to-pt wireless link to a dynamic IP 20Mbps circuit (and its
FiOS Actiontec MI424 router hosting DHCP for a wired and wireless
192.168.1.X network segment) to support the firewall, switches and
non-DHCP 192.168.168.X private network presently behind the DSL modem.

The 'bridge' signal path is not a path I am able to wire with Cat5
ethernet; it is an approximate 100 ft line of sight distance thru three
floors of older construction. At present, I can access the Actiontec's
existing WiFi b/g AP signal with a laptop, getting an adequate signal
and better throughput than the degraded DSL circuit offers.

What I am thinking of is installing a 'wireless bridge' between the FiOS
Actiontec ethernet WAN port and the former DSL network firewall WAN
port, using 600mw WiFi-b/g equipment.

Is this possible? Is this practical?

Thanks.

On Sat, 16 Jul 2011 06:42:09 -0400, News <(E-Mail Removed)> wrote:

>The 'bridge' signal path is not a path I am able to wire with Cat5
>ethernet; it is an approximate 100 ft line of sight distance thru three
>floors of older construction.

Construction is usually described by the materials used, not the age.
Line of sight means you can see the other end. Since you can't see
the other end through 3 floors, you do NOT have line of sight.

I'm also wondering about these 3 floors. Floors are usually 10 to 15
ft seperation. Unless you're shooting at an angle, 100ft would be 8
to 10 floors.

Make my life easy and kindly describe your situation clearly.

If the 3 floors are wood, with not much else in between, you have a
chance. If the 3 floors are concrete, give up now while you're still
sane. Other possibilities will have varying levels of attenuation. In
my experience, 100ft through 3 floors at an angle isn't going to work
because of all the inside junk it has to penetrate. At an angle, it
also has to pass through walls, which often have aluminum foil backed
insulation inside.

>At present, I can access the Actiontec's
>existing WiFi b/g AP signal with a laptop, getting an adequate signal
>and better throughput than the degraded DSL circuit offers.

Ok, no numbers. In order to get 20Mbits/sec thruput, you'll probably
need a 48 or 54Mbits/sec wireless association. That's difficult to do
even in a single room, much less through 3+ floors or walls. Check
your wireless connection (association) speed. I don't think it can be
done.

>What I am thinking of is installing a 'wireless bridge' between the FiOS
>Actiontec ethernet WAN port and the former DSL network firewall WAN
>port, using 600mw WiFi-b/g equipment.

Typical consumer wireless bridges run about 100 mw output. 600mw will
get you perhaps twice the range. For such power to be effective, it
needs to run 600mw at both ends of the wireless line. Otherwise, you
create an "alligator" (an animal with a big mouth and small ears)
which creates more RF pollution.

>Is this possible? Is this practical?

I don't think it's possible. You might be able to get a connection at
some really slow speed (i.e. 1Mbit/sec) but not at 48/54Mbits/sec
association.


--
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

In <(E-Mail Removed)> Jeff Liebermann <(E-Mail Removed)> writes:

>If the 3 floors are wood, with not much else in between, you have a
>chance. If the 3 floors are concrete, give up now while you're still
>sane. Other possibilities will have varying levels of attenuation. In
>my experience, 100ft through 3 floors at an angle isn't going to work
>because of all the inside junk it has to penetrate. At an angle, it
>also has to pass through walls, which often have aluminum foil backed
>insulation inside.

Of course there's always the ethernet-powerline-ethernet solution.

A matched pair of consumer grade units will range from $75 to $200
or thereabouts.

one typical unit:

http://www.staples.com/Netgear-XEB10...product_782584


--
__________________________________________________ ___
Knowledge may be power, but communications is the key
(E-Mail Removed)
[to foil spammers, my address has been double rot-13 encoded]

In article <ivsnmp$e7j$(E-Mail Removed)>,
danny burstein <(E-Mail Removed)> wrote:

> Of course there's always the ethernet-powerline-ethernet solution.
>
> one typical unit:
>
> http://www.staples.com/Netgear-XEB10...t-Switch-Kit/p
> roduct_782584

Promo video is watch-able and well-done. Is the product itself equally
good?

In <siegman-(E-Mail Removed)> AES <(E-Mail Removed)> writes:

>In article <ivsnmp$e7j$(E-Mail Removed)>,
> danny burstein <(E-Mail Removed)> wrote:

>> Of course there's always the ethernet-powerline-ethernet solution.
>>
>> one typical unit:
>>
>> http://www.staples.com/Netgear-XEB10...t-Switch-Kit/p
>> roduct_782584

>Promo video is watch-able and well-done. Is the product itself equally
>good?

Can't speak about that specific unit, but the general concept
works pretty well.

I used a mumble mumble Linksys set three years ago to handle
the connection link from my TiVo to the cable modem. (I've since
picked up TiVo's own wireless unit)

The speed was adequate to not only handle the daily updates, but
worked ok for the video downloads.


--
__________________________________________________ ___
Knowledge may be power, but communications is the key
(E-Mail Removed)
[to foil spammers, my address has been double rot-13 encoded]

On 7/16/2011 12:10 PM, Jeff Liebermann wrote:
> On Sat, 16 Jul 2011 06:42:09 -0400, News<(E-Mail Removed)> wrote:
>
>> The 'bridge' signal path is not a path I am able to wire with Cat5
>> ethernet; it is an approximate 100 ft line of sight distance thru three
>> floors of older construction.
>
> Construction is usually described by the materials used, not the age.
> Line of sight means you can see the other end. Since you can't see
> the other end through 3 floors, you do NOT have line of sight.
>
> I'm also wondering about these 3 floors. Floors are usually 10 to 15
> ft seperation. Unless you're shooting at an angle, 100ft would be 8
> to 10 floors.
>
> Make my life easy and kindly describe your situation clearly.
>
> If the 3 floors are wood, with not much else in between, you have a
> chance. If the 3 floors are concrete, give up now while you're still
> sane. Other possibilities will have varying levels of attenuation. In
> my experience, 100ft through 3 floors at an angle isn't going to work
> because of all the inside junk it has to penetrate. At an angle, it
> also has to pass through walls, which often have aluminum foil backed
> insulation inside.
>

80 feet across, 30 feet up through wood floors and wood lath wall
construction. No metal in either.

Less than 100 feet slant distance; but right, not "LOS".

>> At present, I can access the Actiontec's
>> existing WiFi b/g AP signal with a laptop, getting an adequate signal
>> and better throughput than the degraded DSL circuit offers.
>
> Ok, no numbers. In order to get 20Mbits/sec thruput, you'll probably
> need a 48 or 54Mbits/sec wireless association. That's difficult to do
> even in a single room, much less through 3+ floors or walls. Check
> your wireless connection (association) speed. I don't think it can be
> done.
>

OK, for numbers, I presently get 48Mbps association between the Thinkpad
laptop 3x2 antenna and the Actiontec. Pretty reasonable, actually.


>> What I am thinking of is installing a 'wireless bridge' between the FiOS
>> Actiontec ethernet WAN port and the former DSL network firewall WAN
>> port, using 600mw WiFi-b/g equipment.
>
> Typical consumer wireless bridges run about 100 mw output. 600mw will
> get you perhaps twice the range. For such power to be effective, it
> needs to run 600mw at both ends of the wireless line. Otherwise, you
> create an "alligator" (an animal with a big mouth and small ears)
> which creates more RF pollution.
>

Thinking Engenius EOC-2611P or EAP-3660, both ends @ 600mw.

>> Is this possible? Is this practical?
>
> I don't think it's possible. You might be able to get a connection at
> some really slow speed (i.e. 1Mbit/sec) but not at 48/54Mbits/sec
> association.
>
>

On further consideration?

(Can't do powerline; have tried this and find my 200A panel not friendly.)

On 7/16/2011 3:10 PM, danny burstein wrote:
> In<(E-Mail Removed)> Jeff Liebermann<(E-Mail Removed)> writes:
>
>> If the 3 floors are wood, with not much else in between, you have a
>> chance. If the 3 floors are concrete, give up now while you're still
>> sane. Other possibilities will have varying levels of attenuation. In
>> my experience, 100ft through 3 floors at an angle isn't going to work
>> because of all the inside junk it has to penetrate. At an angle, it
>> also has to pass through walls, which often have aluminum foil backed
>> insulation inside.
>
> Of course there's always the ethernet-powerline-ethernet solution.
>
> A matched pair of consumer grade units will range from $75 to $200
> or thereabouts.
>
> one typical unit:
>
> http://www.staples.com/Netgear-XEB10...product_782584
>
>

Unfortunately, my 200A panel isn't friendly to powerline ethernet
throughput.

In <(E-Mail Removed)> News <(E-Mail Removed)> writes:

[snipppp, regarding powerline ethernet]

>Unfortunately, my 200A panel isn't friendly to powerline ethernet
>throughput.

Unless it's got some wierd type of grounding, surge protections,
or other separations, there shouldn't be an issue.

You might, emphasize might, have to make sure both units
are on the same "leg"s, (either a 50/50 or a 33/66 chance
depending on the input), but that should be the only complication


--
__________________________________________________ ___
Knowledge may be power, but communications is the key
(E-Mail Removed)
[to foil spammers, my address has been double rot-13 encoded]

On 7/16/2011 5:22 PM, danny burstein wrote:
> In<(E-Mail Removed)> News<(E-Mail Removed)> writes:
>
> [snipppp, regarding powerline ethernet]
>
>> Unfortunately, my 200A panel isn't friendly to powerline ethernet
>> throughput.
>
> Unless it's got some wierd type of grounding, surge protections,
> or other separations, there shouldn't be an issue.
>
> You might, emphasize might, have to make sure both units
> are on the same "leg"s, (either a 50/50 or a 33/66 chance
> depending on the input), but that should be the only complication
>
>


Whatever the problem or odds are, they are terminal.

On Sat, 16 Jul 2011 16:13:58 -0400, News <(E-Mail Removed)> wrote:

>80 feet across, 30 feet up through wood floors and wood lath wall
>construction. No metal in either.

Is it just lath and wall board, or lath and *PLASTER*? Anything with
water inside is difficult to penetrate. I ran wireless in the SCZ
City Veteran's hall, which was all lath and plaster on the inside
walls. It was totally impossible to penetrate even one wall, much
less 3 walls. However, it worked well down the hallways, and would
penetrate the wood doors easily. So, I stuffed the place with access
points and antennas. It works, but not through the walls. However,
if it's just wallboard, you have a chance of making this work.

If it were not for the slant path, I might suggest going through the
outside of the building. I've run CAT5 out windows and back in again
in desperation. Wi-Fi will also exit through one window, bounce off
the neighborhing building, and re-enter through another window.
However, with a slant path, forget it.

>Less than 100 feet slant distance; but right, not "LOS".

Yep. It's not LOS.

>OK, for numbers, I presently get 48Mbps association between the Thinkpad
>laptop 3x2 antenna and the Actiontec. Pretty reasonable, actually.

Impressive. Is that while moving traffic, or just at idle? 802.11
tends to slow down as the errors increase, and then speed up when the
traffic stops. Measure your thruput with Jperf (or Iperf) and see
what the wireless link can really do.
<http://openmaniak.com/iperf.php>
<http://ssli.ee.washington.edu/courses/ee299/labs/How_to_Jperf.pdf>
You'll need a 2nd machine to act as a server. Use TCP speeds, not
UDP. Bug me if you need setup help.

Another really simple test is ping. Just ping the other end of the
link at the usual 1 second intervals. Run it for at least 5 minutes.
Watch the latency numbers. If the numbers are a constant number of
msec, you win. However, if the numbers increase erratically, vary
radically, or show no response, then you are experiencing packet loss.
The added latency is cause by packet retransmissions, which show up as
additional time needed to send/return a ping packet. It will work
with packet loss and increased latency, but you won't get full
throughput. The usual cause is co-channel interference, but it can
also be caused by reflections and varying signal strength.

Here's my wireless path to the neighbors. It's usually fairly stable,
but I intentionally fired up my iPhone which provided the necessary
interference.

C:\>ping -t 192.168.1.105
Pinging 192.168.1.105 with 32 bytes of data:
Reply from 192.168.1.105: bytes=32 time=3ms TTL=64
Reply from 192.168.1.105: bytes=32 time=3ms TTL=64
Reply from 192.168.1.105: bytes=32 time=15ms TTL=64
Reply from 192.168.1.105: bytes=32 time=113ms TTL=64
Reply from 192.168.1.105: bytes=32 time=182ms TTL=64
Reply from 192.168.1.105: bytes=32 time=45ms TTL=64
Reply from 192.168.1.105: bytes=32 time=3ms TTL=64
Reply from 192.168.1.105: bytes=32 time=3ms TTL=64
Reply from 192.168.1.105: bytes=32 time=88ms TTL=64

>Thinking Engenius EOC-2611P or EAP-3660, both ends @ 600mw.

I suggest you look at Ubiquiti products instead.
<http://www.ubnt.com>
I favor devices with external antennas (i.e. Bullet)
<http://www.ubnt.com/bullet>
but the build in antenna variety also works well. If you expect
interference on 2.4, think about 5.7GHz.

>On further consideration?

If you can get a reliable 48Mbit/sec association, while moving
traffic, without any interference, it's possible to extend your
20Mbits/sec service.

>(Can't do powerline; have tried this and find my 200A panel not friendly.)

There are power line networking bridges (i.e. ferrite transformers)
that will take care of that problem. More common problems are
interference from other HomePlug users and interference from motor and
switcher noise.

--
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