Help me interpret these results - Linux - various antenas

If I do a less /proc/net/wireless, I get these results:
ralink RT61/Ubuntu linux

Parabolic Antenna 24Dbi
http://www.sendshop.com.br/produtosd...ProdutoID=2572
link 90
level 207
noise 177

Panel antenna, 12Dbi
http://www.sendshop.com.br/produtosd...ProdutoID=2977
link 68
level 201
noise 177

Big Tin Cantenna ?Dbi
link 52
level 191
noise 161

Small Tin Cantenna ?Dbi
link 59
level 193
noise 161

With this last one, I can still download as fast as my provider allows
(15KBs), but it sometimes errors out on big downloads.
If there is no noise difference between my parabolic and my panel, and
the level is so similar (201 vs 207) why is the "link"so much better
with my parabolic ?

Also, I have the option to decrease the wireless card speed from 54Mbs
right down to 1Mbs. Will this improve reception, and will it affect my
downloads in any way (speed or quality wise ?)
TIA

Shadow <sh@dow> hath wroth:

>If I do a less /proc/net/wireless, I get these results:
>ralink RT61/Ubuntu linux

Those are RSSI values on a scale of 0 to -255. The values in -dBm are
far more useful.

>Parabolic Antenna 24Dbi
>http://www.sendshop.com.br/produtosd...ProdutoID=2572
>link 90
>level 207
>noise 177
>
>Panel antenna, 12Dbi
>http://www.sendshop.com.br/produtosd...ProdutoID=2977
>link 68
>level 201
>noise 177
>
>Big Tin Cantenna ?Dbi
>link 52
>level 191
>noise 161
>
>Small Tin Cantenna ?Dbi
>link 59
>level 193
>noise 161
>
>With this last one, I can still download as fast as my provider allows
>(15KBs), but it sometimes errors out on big downloads.

15KBytes/sec = 96Kbits/sec. Is that all your provider allows? Is
this a satellite or cellular link? The errors can be anything from
interefernce to the ISP throttling your massive download. For
example, Hughes.net FAP (fair access policy).

>If there is no noise difference between my parabolic and my panel, and
>the level is so similar (201 vs 207) why is the "link"so much better
>with my parabolic ?

The "noise" number is the signal level when there is no detectable
data coming in. That's basically between packets. It's a measure of
interference, background noise, thermal noise, and other systems on
adjacent channels. In this case, it's really a measure of how
directional your antenna is in excluding these sources of crud, er...
noise. The narrower antenna beamwidth antenna simply picks up less
crud from the sides of the antenna. However, if you perchance aimed
the antenna as one of these sources of crud, the noise level
indication will increase.

>Also, I have the option to decrease the wireless card speed from 54Mbs
>right down to 1Mbs. Will this improve reception, and will it affect my
>downloads in any way (speed or quality wise ?)

Dropping the speed is normally NOT a function of the wireless card in
an infrastructure network. The speed is usually set by the access
point. You can change the speed at the client, but the access point
still retains control. Some clients allow you to set the maximum
speed. When the client and AP exchange capabilities, it will use only
those speeds that are allowed up to a maximum. I know of several
access points that ignore this "feature" and set the speed to what
they consider acceptable.

In ad-hoc mode, which is all clients, the speed is set by the
individual clients.

Thruput and connection speed are different animals. In general, your
TCP thruput will be half your wireless connnection speed. A
54Mbit/sec "association" will yield about a 25Mbit/sec TCP thruput.
UDP is a bit faster. See:
<http://wireless.wikia.com/wiki/Wi-Fi#Performance_and_Speed>
From the table, note that a cutting the speed in half, results in a
1.4 times increase in range.

You certainly will get more range from the slower speeds than the
faster. That's because the access point tries to maintain a constant
bit error rate. If the error rate climbs, the access point will slow
down the connection. Therefore, if the access point is in control of
the speed, the signal "quality" is contant at any speed.

This might be worth reading. It's a copy of the Intel Wireless
Hotspot Deployment Guide, which is currently missing from the Intel
web pile. Don't tell Intel that I'm posting it without their
permission:
<http://802.11junk.com/jeffl/crud/Intel%20HotSpot%20Guide.pdf>
It covers quite a bit of what you're asking, in addition to much of
the basics of wi-fi and how it works. 1.8MBytes.

--
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 Thu, 23 Aug 2007 16:50:59 -0700, Jeff Liebermann <(E-Mail Removed)>
wrote:
>> Don't tell Intel that I'm posting it without their
>permission:
Let it be a secret between me, you and "the Internet users"

><http://802.11junk.com/jeffl/crud/Intel%20HotSpot%20Guide.pdf>
Will read (downloaded, but I'm too tired atm ... thanks)
>It covers quite a bit of what you're asking, in addition to much of
>the basics of wi-fi and how it works. 1.8MBytes.

On Thu, 23 Aug 2007 16:50:59 -0700, Jeff Liebermann <(E-Mail Removed)>
wrote:

>Shadow <sh@dow> hath wroth:
>
>>If I do a less /proc/net/wireless, I get these results:
>>ralink RT61/Ubuntu linux
>
>Those are RSSI values on a scale of 0 to -255. The values in -dBm are
>far more useful.
Is there any kind of rough comparison I can make, using these
figures ?
>
>>Parabolic Antenna 24Dbi
>>http://www.sendshop.com.br/produtosd...ProdutoID=2572
>>link 90
>>level 207
>>noise 177
>>
>>Small Tin Cantenna ?Dbi
>>link 59
>>level 193
>>noise 161
>>
>>With this last one, I can still download as fast as my provider allows
>>(15KBs), but it sometimes errors out on big downloads.
From the pdf you thought I might illegally download, I got
these figures:
At 1Mbps = range 350 feet
At 54 Mbps = range 90 feet
So I could maybe get pretty good reception with my small can,
if I turn down the speed ? It would be as if my provider antenna was
1/4 the distance away ?
>
>15KBytes/sec = 96Kbits/sec. Is that all your provider allows?
Yes
> Is this a satellite or cellular link?
Its a Brazilian monopoly. It's that or a 56kbps modem.
> The errors can be anything from
>interefernce to the ISP throttling your massive download. For
>example, Hughes.net FAP (fair access policy).
>
>>If there is no noise difference between my parabolic and my panel, and
>>the level is so similar (201 vs 207) why is the "link"so much better
>>with my parabolic ?
>
>The "noise" number is the signal level when there is no detectable
>data coming in. That's basically between packets. It's a measure of
>interference, background noise, thermal noise, and other systems on
>adjacent channels. In this case, it's really a measure of how
>directional your antenna is in excluding these sources of crud, er...
>noise. The narrower antenna beamwidth antenna simply picks up less
>crud from the sides of the antenna. However, if you perchance aimed
>the antenna as one of these sources of crud, the noise level
>indication will increase.
Which is why my small tin cantenna has less trash, it's more
directional. OK.
>
>>Also, I have the option to decrease the wireless card speed from 54Mbs
>>right down to 1Mbs. Will this improve reception, and will it affect my
>>downloads in any way (speed or quality wise ?)
>
>Dropping the speed is normally NOT a function of the wireless card in
>an infrastructure network. The speed is usually set by the access
>point. You can change the speed at the client, but the access point
>still retains control. Some clients allow you to set the maximum
>speed. When the client and AP exchange capabilities, it will use only
>those speeds that are allowed up to a maximum. I know of several
>access points that ignore this "feature" and set the speed to what
>they consider acceptable.
I can set my card to 1MB using linux commands, or my RutilT
utility. I can even make it pretend it's a 802.11b card.

http://rt2x00.serialmonkey.com/wiki/index.php/Downloads

>
>In ad-hoc mode, which is all clients, the speed is set by the
>individual clients.
>
>Thruput and connection speed are different animals. In general, your
>TCP thruput will be half your wireless connnection speed. A
>54Mbit/sec "association" will yield about a 25Mbit/sec TCP thruput.
>UDP is a bit faster. See:
><http://wireless.wikia.com/wiki/Wi-Fi#Performance_and_Speed>
>From the table, note that a cutting the speed in half, results in a
>1.4 times increase in range.
>
>You certainly will get more range from the slower speeds than the
>faster.
I will default it to 1Mbps then....
>That's because the access point tries to maintain a constant
>bit error rate. If the error rate climbs, the access point will slow
>down the connection. Therefore, if the access point is in control of
>the speed, the signal "quality" is contant at any speed.
The provider runs a Zinwell ZWA-G120 with an omni antenna set
about 1km from my house. I have a clear sight to it.

Shadow <sh@dow> hath wroth:

>>>If I do a less /proc/net/wireless, I get these results:
>>>ralink RT61/Ubuntu linux
>>
>>Those are RSSI values on a scale of 0 to -255. The values in -dBm are
>>far more useful.

> Is there any kind of rough comparison I can make, using these
>figures ?

Yes, but I would need to get the data sheet on the chipset in your
client radio. Many of the manufacturers have RSSI to signal level (in
-dBm) conversion tables or formula. It's a common issue with all
wireless products including cellular handsets. For exmple:
<http://www.maxstream.net/support/knowledgebase/article.php?kb=65>

<http://en.wikipedia.org/wiki/RSSI>
<http://en.wikipedia.org/wiki/RCPI>

I'm lazy. Find the FCC ID on your Ralink RT61 and look up the test
report on the FCC ID web pile. That will give you the radio chipset
used. Then go to the manufacturers web pile, and get the data sheet
on the chip. That should give you a clue as to how to interpret the
RSSI or RCPI values.

> From the pdf you thought I might illegally download, I got
>these figures:
> At 1Mbps = range 350 feet
> At 54 Mbps = range 90 feet

The figures from the pilphered Intel document are a bit optimistic for
indoor wireless range. It's fairly close to what I would expect with
really good radios, really nice antennas, no interference, and a
perfect line of sight.

> So I could maybe get pretty good reception with my small can,
>if I turn down the speed ? It would be as if my provider antenna was
>1/4 the distance away ?

Exactly. However, there's no free lunch. 1Mbit/sec uses about 60
times more "air time" as the same amount of data at 54Mbits/sec. That
means there's little air time left of other users on the same channel.
If they have a tweaked MAC layer (very common) that treats 802.11b
signals as interference, the chances of clobbering your ultra long
packet time is quite high. 802.11 will fragment the packets to
improve the odds, but then your thruput will drop even more.
Therefore, if you're going to trade speed for range, I suggest you
limit yourself to using 802.11g (OFDM) speeds instead.

Incidentally, I run several systems with the AP's locked to
12Mbits/sec wireless speeds. It's slower, but more reliable (fewer
retransmissions). When the internet speed is limited to 1Mbit/sec, it
makes no sense to go 54Mbits/sec (25Mbit/sec TCP thruput) when slower
will work exactly the same from the users perspective.

> Which is why my small tin cantenna has less trash, it's more
>directional. OK.

Cantennas are not that directional. See:
<http://802.11junk.com/jeffl/antennas/coffee2400/index.html>
I would say a -3dB bandwidth of 100 degrees. Compare that with a
biquad of 50 degrees, or a 24dBi dish of about 7 degrees.

> I can set my card to 1MB using linux commands, or my RutilT
>utility. I can even make it pretend it's a 802.11b card.
>http://rt2x00.serialmonkey.com/wiki/index.php/Downloads

I would use 802.11b speeds only if desperate. 802.11g is far more
resistant to reflections and interference.

> The provider runs a Zinwell ZWA-G120 with an omni antenna set
>about 1km from my house. I have a clear sight to it.

Try to guess the gain of the omni antenna by it's length. If there's
any coax cable involved, guess the size, length, and attenuation. Then
go unto:
<http://wireless.wikia.com/wiki/Wi-Fi#Link_Calculations>
and calculate at what speed you can still have at least 10dB SOM (fade
margin) at 1km. I've posted numerous walk through on how to do the
calcs in this newsgroup. That will tell you how much antenna gain
you'll need.

--
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