Locating interference to WIFI network

A relative of mine lives in an apartment building and sometimes his
WIFI connection drops.
He's no longer had non-WIFI friendly cordless phones. Using netstumbler
to scan, he has moved to the least busy channel. He still gets dropped
off of WIFI.

While maybe this is caused by the WIFI AP which is on the same channel,
but rarely shows up, I'm trying to pinpoint things more.

At Interop, I saw Cognio's product
http://www.cognio.com/solutions_mobile.html
which not only does spectrum analysis, the software knows what a
cordless phone or microwave oven look like, so it automatically tells
you what type of interference there is and then uses gieger counter
like features to help you pinpoint the location. Well, I'm not spending
$5000 to figure out this issue.

Is there an inexpensive way to do this?
I read some old postings which mention using a Proxim 7200 or 7400 card
or even a Teletronics WL2000. Even if I could get one - I'm not sure
I'd be able to read the spectrum analyzer results to figure out what
the interference is... Is there any tutorial on how to read these
things properly? Also, dropped WIFI doesn't happen all the time (who
knows if there interferece is there all the time) --- I would really
need to know what event happened at that point. Also, as I mentioned,
channels 1, 6, 11 are all taken (although 1 is far enough, that it
rarely is registered by his WIFI router).

Any other recommendations?

Thanks

(E-Mail Removed) wrote:
>A relative of mine lives in an apartment building and sometimes his
>WIFI connection drops.

Yeah, unfortunately, even if he knew what the problem was, hemiht not
be able to fix it. Rather difficult to ask your neighbors not to use
their microwave, cordless phones, etc, eh?

You could also have networks with SSID broadcast turned off
interfering. I've seen them with the Auditor Security Collection, but
if, as you say, it's intermittent, it may be hard to find.

Is his apartment small enough for him to convert to 802.11a? AFAICT
that won't penetrate walls, so he shouldn't have any interference
issues...

William P.N. Smith wrote:
> (E-Mail Removed) wrote:
> >A relative of mine lives in an apartment building and sometimes his
> >WIFI connection drops.
>
> Yeah, unfortunately, even if he knew what the problem was, hemiht not
> be able to fix it. Rather difficult to ask your neighbors not to use
> their microwave, cordless phones, etc, eh?

Well, part of the thing is to be able to confirm that it IS something
outside of his apartment interfering. If it happens to be something in
his apartment, which we didn't think about, he could control it...
Otherwise, I'll tell him to put up RF shielding wallpaper or paint :-)
I don't think he'll go to 802.11a at this point at least.

On 15 Dec 2005 15:17:08 -0800, (E-Mail Removed) wrote:

>A relative of mine lives in an apartment building and sometimes his
>WIFI connection drops.

Could you offer some clue as to how often is "sometimes". Getting
dropped a few times a day is normal in an RF congested environment. I
know one network that goes down around lunch time due to the large
number of leaky microwave ovens operating in a nearby cafeteria.

>He's no longer had non-WIFI friendly cordless phones. Using netstumbler
>to scan, he has moved to the least busy channel. He still gets dropped
>off of WIFI.

What channel did he pick? The non-overlapping channels are 1, 6, and
11. If you pick a channel in between such as 3, you will get
interference from users on *BOTH* channels 1 and 6. Stay with 1, 6,
and 11 even if occupied.

>While maybe this is caused by the WIFI AP which is on the same channel,
>but rarely shows up, I'm trying to pinpoint things more.

An AP on the same channel will not cause a connection to drop. It may
slow down, or become erratic, but it will not disconnect. Wi-Fi and
spread spectrum is fairly resistant to co-channel interference.
Incidentally, 802.11g is far more resistant to disconnects than
802.11b.

Disconnects are usually caused by non-802.11 type of interference such
as cordless phones, microwave ovents, 2.4GHz security cameras and
other sources of 2.4Ghz junk that tend to be continuously
transmitting. Bluetooth will cause disconnects, but only if located
almost on top of the access point or client radio.

>At Interop, I saw Cognio's product
>http://www.cognio.com/solutions_mobile.html
>which not only does spectrum analysis, the software knows what a
>cordless phone or microwave oven look like, so it automatically tells
>you what type of interference there is and then uses gieger counter
>like features to help you pinpoint the location. Well, I'm not spending
>$5000 to figure out this issue.

Ouch. See:
http://www.bvsystems.com/Products/WLAN/WLAN.htm
About $1500 for the YellowJacket. Be sure to view the video clips at
the bottom of:

http://www.bvsystems.com/Products/WL...wjacket%2B.htm
to see how it looks. What you'll find is that it's NOT easy to
identify the exact source of interference with one of these.

>Is there an inexpensive way to do this?

Borrow or rent something. The Proxim 7400 based system is VERY slow
and insensitive. I use several but have largely given up in favor of
using an MMDS downconverter and a portable (borrowed) spectrum
analyzer.

>I read some old postings which mention using a Proxim 7200 or 7400 card
>or even a Teletronics WL2000. Even if I could get one - I'm not sure
>I'd be able to read the spectrum analyzer results to figure out what
>the interference is... Is there any tutorial on how to read these
>things properly?

Not really. It's best to do this by example. Put the device near a
known source of interference and see what the display looks like. Even
so, there is a considerable learning curve and all too much guesswork.
I can do it with a decent spectrum analyzer, but even with years of
experience, I'm still guessing all too often. There's also the
problem of locating the source of interference. Big directional dish
antennas are available, but reflections make using them also an art.
I was once trying to locate an 802.11 user that had an obvious high
power amplifier that was causing considerable interference. The
signal was so strong that I spent most of the day chaseing reflections
instead of the source.

>Also, dropped WIFI doesn't happen all the time (who
>knows if there interferece is there all the time) --- I would really
>need to know what event happened at that point. Also, as I mentioned,
>channels 1, 6, 11 are all taken (although 1 is far enough, that it
>rarely is registered by his WIFI router).
>
>Any other recommendations?
>
>Thanks
--
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

Thanks - read the embedded reply:

Jeff Liebermann wrote:
> On 15 Dec 2005 15:17:08 -0800, (E-Mail Removed) wrote:
>
> >A relative of mine lives in an apartment building and sometimes his
> >WIFI connection drops.
>
> Could you offer some clue as to how often is "sometimes".

I'm not exactly sure... In the three or so hours people were in front
of the computer, it happened twice.


> >He's no longer had non-WIFI friendly cordless phones. Using netstumbler
> >to scan, he has moved to the least busy channel. He still gets dropped
> >off of WIFI.
>
> What channel did he pick? The non-overlapping channels are 1, 6, and
> 11. If you pick a channel in between such as 3, you will get
> interference from users on *BOTH* channels 1 and 6. Stay with 1, 6,
> and 11 even if occupied.

Yah, I moved him from 11 (which had a a nearby AP, which was getting
received as 85% signal strength) to channel 1, which had an AP which is
only sometimes detectable, and comes in at around 60%.


> >While maybe this is caused by the WIFI AP which is on the same channel,
> >but rarely shows up, I'm trying to pinpoint things more.
>
> An AP on the same channel will not cause a connection to drop. It may
> slow down, or become erratic, but it will not disconnect. Wi-Fi and
> spread spectrum is fairly resistant to co-channel interference.
> Incidentally, 802.11g is far more resistant to disconnects than
> 802.11b.
Would I be better on a channel with a stronger competing signal which
is always there -- or a weak signal which the PC only sometimes picks
up? That is, does the WIFI 'learn' how to get around the stronger
signal... but get 'interrupted' by a signal which comes and goes?


> >At Interop, I saw Cognio's product
>>....
> >$5000 to figure out this issue.
>
> Ouch. See:
But, it would take out alot of the guesswork and the trial and error...
But, I'm still not paying $5000, unless I'd be doing it professionally.

> http://www.bvsystems.com/Products/WLAN/WLAN.htm
> About $1500 for the YellowJacket. Be sure to view the video clips at
> the bottom of:
Yep, I saw those at Interop as well.... It's still not worth $1500 for
this purpose...

> >Is there an inexpensive way to do this?
>
> Borrow or rent something.

Any idea where to go to rent something like that? Again, since it's has
a large learning curve, it may not even be too useful for the amount of
time I'd be renting it.


>The Proxim 7400 based system is VERY slow
> and insensitive.
In the meantime, I ordered one of those FHSS cards, to see if it helps
at all...

> >the interference is... Is there any tutorial on how to read these
> >things properly?
>
> Not really. It's best to do this by example.
I've seen some presentations on the web, also... hopefully they'll be
of some help.

> There's also the
> problem of locating the source of interference.
I'm also an SWLer... and I know that there are hobbiests (probably
mostly HAMs) who go on signal hunting (not WIFI signals) trips for
fun...

Thanks again...

On 17 Dec 2005 21:12:13 -0800, (E-Mail Removed) wrote:

>> Could you offer some clue as to how often is "sometimes".
>
>I'm not exactly sure... In the three or so hours people were in front
>of the computer, it happened twice.

Ok. About once an hour or more. That's not unusual. I would
guess(tm) a microwave oven. It's not another wireless LAN because
they tend to get used for longer periods of time. I can deduce quite
a bit from the interference pattern and timing. Anything else you can
supply such as the duration of the outage, whether it constitutes a
disconnect or a loss in download speed, and whether it coincides with
commercial breaks on TV.

>Yah, I moved him from 11 (which had a a nearby AP, which was getting
>received as 85% signal strength) to channel 1, which had an AP which is
>only sometimes detectable, and comes in at around 60%.

The problem is always the hidden access points that don't broadcast
their SSID. Your access point search utility will not show these. I
suggest you download and run one of the Linux LiveCD's that has Kismet
on it and search for other sources of 802.11 interference.
http://www.remote-exploit.org/index.php/Auditor_main
It won't show non-802.11 interference but at least you can do an
effective job of selecting the channel.

>Would I be better on a channel with a stronger competing signal which
>is always there -- or a weak signal which the PC only sometimes picks
>up?

Good questions. Spread spectrum shows up as valid data that can be
demodulated. The listen before talk system will wait for an empty
time slot and interleave the transmit signals. I've run about 12
access points simultaneously in the same room copying files furiously
without disconnects. Traffic did slow down considerably but never
stopped.

The only difference between interference from an on channel SS signal
and one on the adjacent channel is the signal levels. The
demodulation process is not synchronized with the carrier so it really
doesn't matter what frequency (channel) the interference is on.

Therefore, the channel weakest interference source is the best. What
most people don't bother to do is check the channels on either side of
the selected channel. A strong signal one or two channels on either
side of the selected channel can create almost as much interference as
a signal on the selected channel. It's not enough to just check
channels 1, 6, and 11. You also have to check the nearby channels.

You can sometimes "see" interference if your client radio has a
"noise" level indicator. It won't tell you anything other than
there's some source of noise nearby, but it does help select the
channel. Pick the one with the lowest "noise" level.

>That is, does the WIFI 'learn' how to get around the stronger
>signal... but get 'interrupted' by a signal which comes and goes?

Sorta. Recent tweaks to Part 15.247 allow limited adaptive frequency
selection. Many new access points have an automatic setting instead
of a fixed channel selection (Netgear AutoCell).
http://www.autocell.com
Allegedly, they find the "best" channel, usually by just looking for
the lowest noise level and possibly other users. When they select a
new channel, the client radios will follow. This feature sounds good,
but I'm not convinced it does anything more than present a moving
target for users trying to find an optimum channel selection. I
haven't played with AutoCell as it requires an additional license.

>> http://www.bvsystems.com/Products/WLAN/WLAN.htm
>> About $1500 for the YellowJacket.

>Yep, I saw those at Interop as well.... It's still not worth $1500 for
>this purpose...

Find someone that has one and borrow it. I've thought of buying one
and renting it to the local hacks. There seems to be a market.

>Any idea where to go to rent something like that? Again, since it's has
>a large learning curve, it may not even be too useful for the amount of
>time I'd be renting it.

Nope. What country are you in? What city are you in? There should
be a local test equipment rental house in all the major cities. The
problem is they tend to charge 10-15% of the purchase price per month
for rentals. I don't think you'll like the cost.

Find the local WISP (Wireless Internet Service Provider) and see if
they have some test gear. Same with the local commercial 2-way radio
shop. If the municipal radio shop is friendly, they might have
something useful. I've had best luck with the few radio station
engineers that actually work on the hardware. They always seem to
have decent test equipment. They may not loan it to you, but you
might be able to bribe them into doing the sniffing (site survey) for
you. Last resort are the local hams:
http://www.eham.net/forums/FoxHunting
Like me, they tend to have workable junk, some expertise, and a
willingness to help.

>>The Proxim 7400 based system is VERY slow
>> and insensitive.

>In the meantime, I ordered one of those FHSS cards, to see if it helps
>at all...

Here's a PowerPoint presentation on the card from the North Texas
Microwave Society:
http://www.ntms.org/files/$20%20WIFI%20SA.zip
I don't know how they got the signal strength up that high as shown on
the "snoop" program display. I have 6 cards and none of them ever
show anything stronger than perhaps half scale. Maybe I have 6 dead
cards?

>I'm also an SWLer... and I know that there are hobbiests (probably
>mostly HAMs) who go on signal hunting (not WIFI signals) trips for
>fun...

I used to do that all the time on VHF long ago. We still have
transmitter hunts locally, usually when someone has their transmitter
stuck on the air and we have to go find the culprit. I've been
(slowly) writing an article on how to do direction finding using the
audio delay through a repeater.

--
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

On 17 Dec 2005 21:12:13 -0800, (E-Mail Removed) wrote:

>>The Proxim 7400 based system is VERY slow
>> and insensitive.
>In the meantime, I ordered one of those FHSS cards, to see if it helps
>at all...

Note that the Proxim 7400 driver does not work with XP, but works well
with previous Windoze mutations.

--
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

Thanks. Again, replies inline:

Jeff Liebermann wrote:
> On 17 Dec 2005 21:12:13 -0800, (E-Mail Removed) wrote:

> The problem is always the hidden access points that don't broadcast
> their SSID.

Yah, I knew that some software displayed the SSID for those who don't
broadcast it, I just didn't know which. (Which is the reason why they
say that not broadcasting SSID, doesn't really help security much.)

> a signal on the selected channel. It's not enough to just check
> channels 1, 6, and 11. You also have to check the nearby channels.

The software I was using showed WIFI activity in all channels.

> Find the local WISP (Wireless Internet Service Provider) and see if
> they have some test gear.
Hopefully I'll be able to google for one of those in the area...

> I've had best luck with the few radio station
> engineers that actually work on the hardware.

My radio engineer contacts are about 4000 miles away :-)
I'll try to contact some Ham/SWL people I deal with and see if they
have any ideas (since none are local - even if they have the equipment.
My relative is in Southern CT and I'm in the greater NYC area.


>
> >>The Proxim 7400 based system is VERY slow
> >> and insensitive.
>
> >In the meantime, I ordered one of those FHSS cards, to see if it helps
> >at all...
Yep, I know it's not XP compatible...


Thanks again! I'll try to check up on the possibilities you
mentioned....

Jeff Liebermann <(E-Mail Removed)> wrote:
> (E-Mail Removed) wrote:
[Jeff wrote]
>>>The Proxim 7400 based system is VERY slow
>>> and insensitive.

>>In the meantime, I ordered one of those FHSS cards, to see if it helps
>>at all...

>Note that the Proxim 7400 driver does not work with XP, but works well
>with previous Windoze mutations.

Yeah, fortunately, I discovered an old Win2K laptop when my card came
in. Unfortunately, I discovered that Jeff's right (imagine my
surprise!), and it's slow enough to be essentially unusable, and deaf
enough to only see APs right next to it. As a spectrum analyzer to be
used as an interference location tool it's not very useful.

A real spectrum analyzer with decent speeds and high sensitivity is,
as we're seeing, a very expensive piece of electronics, and even in
trained hands can only point out that there's interference. Doing
something about it in the 2.4GHz band is another rathole...

On Sun, 18 Dec 2005 11:47:26 -0500, William P.N. Smith
<(E-Mail Removed)> wrote:

>Jeff Liebermann <(E-Mail Removed)> wrote:
>> (E-Mail Removed) wrote:
>[Jeff wrote]
>>>>The Proxim 7400 based system is VERY slow
>>>> and insensitive.
>
>>>In the meantime, I ordered one of those FHSS cards, to see if it helps
>>>at all...
>
>>Note that the Proxim 7400 driver does not work with XP, but works well
>>with previous Windoze mutations.

I forgot that I had made an AVI file of what mine looks like scanning
for my access point about 10 feet away.
http://802.11junk.com/jeffl/sa/Proxim7400-01.avi (2.5MB)
My appologies for the otto focus, shakey picture (no tripod), and
oversize video clip.

>Yeah, fortunately, I discovered an old Win2K laptop when my card came
>in. Unfortunately, I discovered that Jeff's right (imagine my
>surprise!), and it's slow enough to be essentially unusable, and deaf
>enough to only see APs right next to it. As a spectrum analyzer to be
>used as an interference location tool it's not very useful.

Well, I cheated a bit. I also have a Symbol PPT4340 which has
essentially the same card inside and came with similar software. It
too is deaf and slow. However, I added a 2.4Ghz RF amplifier and an
antenna connector, making it somewhat more useful. What bothers me is
that my signal levels never seem to go above half scale on all my
cards and laptops that I've tried running it. Even with an RF
amplifier and right next to the access point, it only goes half scale.
Yet, the photos on various web pages show almost full scale. Either I
have 5 broken cards, or I'm doing something wrong.

>A real spectrum analyzer with decent speeds and high sensitivity is,
>as we're seeing, a very expensive piece of electronics, and even in
>trained hands can only point out that there's interference. Doing
>something about it in the 2.4GHz band is another rathole...

That depends on what you consider expensive. I've been tempted to buy
something useful and rent it to my competitors. There's definitely a
need for interference identification. What I've been doing is using a
modified MMDS down-converter and a borrowed Tektronix low frequency
spectrum analyzer. Some of the Pacific Monolithics and Conifer dish
antennas came with MMDS down-converters built into the antenna feed.
http://apache.airnet.com.au/~fastinfo/wireless/PacMon/
Instead of destroying the MMDS converter and replacing the feed as
shown above, I rip out the 2.4GHz filter, supply power through the
coax, and watch the spectra on a lower frequency spectrum analyzer.
http://i.cmpnet.com/csd/gifs/9804fe11.gif
At 2.400 to 2.485 Ghz, the IF appears from 122 to 207Mhz, which is
easily viewable on a cheaper spectrum analyzer.

The MMDS downconverter is currently mounted behind a dish which often
lives on top of a 30ft telescoping fiberglass pole used for site
surveys. When I do another one, I'll try to remember to take photos
before I get arrested for being a theat to homeland security by
walking around with such a contrivance.

--
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