signal numbers off WPC54G make sense?


I'm wondering if these numbers mean anything to you guys in
the know about wireless stuff.
With a standard antenna on my WAP54G I see these numbers on
the screen of a laptop connected to the AP using a WPC54G PC
Card: noise -68dBm, signal -78dBm, link quality (I think
it's called) 27%.
Then I put a small corner reflector on one of the WAP54G
antennas and it bumped up the lightbars and I got
noise -58dBm, signal -68dBm, and quality 47%.
I was surprised at several things. I was a bit surprised
that the S/N was so low to begin but I can accept that. But
why the heck wouldn't the S/N improve with a stronger signal
from the AP? These numbers show both S and N going up 10dB.
Finally, especially at low S/N I would have expected any
measure of data quality would do more than double if S/N
went up 10dB.
Clearly I'm not up on this wireless system. I have looked
around at some technical places but have never seen this
subject discussed. Could you guys in the know point me where
I can make sense of all this? Thanks.
Reply to
rob
Loading thread data ...
rob hath wroth:
Hang on a minute. I gotta put on my wizards hat.
I think you have the signal and noise backwards. The noise cannot be more than the signal or you wouldn't hear anything. Assuming your numbers are backwards, that yields:
signal = -68dBm noise = -78dBm S/N = 10dB (which sucks)
The link quality number varies with the chipset used. There are different algorithms all of which are based on the percentage of the S/N ratio required to get a perfect BER (bit error rate) at whatever speed it's running at. I have the numbers buried here somewhere if you really want them. More commonly, the link quality is a measure of the actual BER. The more receive errors, the worse the link quality. That causes some rather bizarre results in the presense of interference. See below.
Again backwards. I'll assume:
signal = -58dBm noise = -68dBm S/N = 10dB (which sucks)
Note that BOTH the signal and noise levels were increased by the corner reflector. That's what happens when you have a source of interference in the line of sight. The chipset treat everything that is not 802.11 or the desired access point as "noise".
The reason the link quality went from 27% to 47% is that the strong signal made it easier to demodulate the data. In theory, the link quality should exactly follow the S/N ratio, but different types of intereference cause different types of data reception errors.
Please check your numbers and try again?
Nope. My guess(tm) is that your WPC54G uses the data error rate to determine the link quality. You could simply add a noise or interference source and the signal levels will not change. The noise level should change with the added interference. However, if you are at some threshold (as you probably are with such a lousy SNR), then the effects will be proportionally larger due to threhold effects.
Not exactly. It's in the data sheets and application notes for the chipset, which often require an NDA or sacrifice of the first born to obtain. 2nd best are articles on the topic. Try searching:
There are others but these are a good start. If I find anything useful, I'll post.
I sorta blundered across these articles that looks useful:
Reply to
Jeff Liebermann
Believe me, the numbers I gave were what is clearly seen on the screen, checked numerous times. I am really confused as to why the noise level would go up with the signal. I'm wondering if the guy at Linksys that formatted the output was thinking more about lunch than what he was doing.
Thanks for the links, I'll check them and see what I can learn about this business. Rob
Jeff Liebermann wrote:
Reply to
rob
rob hath wroth:
Try the same measurements with: 1. Using the Linksys client instead of Wireless Zero Config 2. Netstumbler 3. A different laptop
It doesn't make any sense to have more noise than signal. Something is weird on your WPC54G client. If you change your setup from Windoze Wireless Zero Config, and switch to the Linksys supplied client, you can get S/N info from the "Link Information" tab, and not get entangled in the % link quality conversion. Hopefully, you'll see better numbers. Also, check that you have the latest WPC54G drivers installed.
> >Thanks for the links, I'll check them and see what I can >learn about this business. Rob > >Jeff Liebermann wrote: > >> rob hath wroth: >> >> >>>I'm wondering if these numbers mean anything to you guys in >>>the know about wireless stuff. >> >> >> Hang on a minute. I gotta put on my wizards hat. >> >> >>>With a standard antenna on my WAP54G I see these numbers on >>>the screen of a laptop connected to the AP using a WPC54G PC >>>Card: noise -68dBm, signal -78dBm, link quality (I think >>>it's called) 27%. >> >> >> I think you have the signal and noise backwards. The noise cannot be >> more than the signal or you wouldn't hear anything. Assuming your >> numbers are backwards, that yields: >> >> signal = -68dBm >> noise = -78dBm >> S/N = 10dB (which sucks) >> >> The link quality number varies with the chipset used. There are >> different algorithms all of which are based on the percentage of the >> S/N ratio required to get a perfect BER (bit error rate) at whatever >> speed it's running at. I have the numbers buried here somewhere if >> you really want them. More commonly, the link quality is a measure of >> the actual BER. The more receive errors, the worse the link quality. >> That causes some rather bizarre results in the presense of >> interference. See below. >> >> >>>Then I put a small corner reflector on one of the WAP54G >>>antennas and it bumped up the lightbars and I got >>>noise -58dBm, signal -68dBm, and quality 47%. >> >> >> Again backwards. I'll assume: >> >> signal = -58dBm >> noise = -68dBm >> S/N = 10dB (which sucks) >> >> Note that BOTH the signal and noise levels were increased by the >> corner reflector. That's what happens when you have a source of >> interference in the line of sight. The chipset treat everything that >> is not 802.11 or the desired access point as "noise". >> >> The reason the link quality went from 27% to 47% is that the strong >> signal made it easier to demodulate the data. In theory, the link >> quality should exactly follow the S/N ratio, but different types of >> intereference cause different types of data reception errors. >> >> >>>I was surprised at several things. I was a bit surprised >>>that the S/N was so low to begin but I can accept that. But >>>why the heck wouldn't the S/N improve with a stronger signal >> >>>from the AP? These numbers show both S and N going up 10dB. >> >> Please check your numbers and try again? >> >> >>>Finally, especially at low S/N I would have expected any >>>measure of data quality would do more than double if S/N >>>went up 10dB. >> >> >> Nope. My guess(tm) is that your WPC54G uses the data error rate to >> determine the link quality. You could simply add a noise or >> interference source and the signal levels will not change. The noise >> level should change with the added interference. However, if you are >> at some threshold (as you probably are with such a lousy SNR), then >> the effects will be proportionally larger due to threhold effects. >> >> >>>Clearly I'm not up on this wireless system. I have looked >>>around at some technical places but have never seen this >>>subject discussed. Could you guys in the know point me where >>>I can make sense of all this? Thanks. >> >> >> Not exactly. It's in the data sheets and application notes for the >> chipset, which often require an NDA or sacrifice of the first born to >> obtain. 2nd best are articles on the topic. Try searching: >> >> >> There are others but these are a good start. If I find anything >> useful, I'll post. >> >> I sorta blundered across these articles that looks useful: >>
>> >> >>
Reply to
Jeff Liebermann
The only setup I know of that I used to configure my PC Card on the laptop was what came with the card. Don't know what Win Zero Config is but I'll try to find out. I have read about netstumbler but it seemed like a lot of stuff to wade into and since I finally got the laptop connecting, I didn't bother with it. I'll look it up again and see what it is. Don't have another laptop.
I have not yet seen any numbers for what S/N these wireless links operate at (and I have been looking thru the links you gave out). But the telemetry links I have played with run at S/N below one to optimize bit rate/power which our friend Shannon pointed out. As I understand it, thruput is linear with BW but logrithmic in S/N.
Thanks
Jeff Liebermann wrote: Try the same measurements with: 1. Using the Linksys client instead of Wireless Zero Config 2. Netstumbler 3. A different laptop
Reply to
rob
On Mon, 05 Mar 2007 12:35:42 -0600, rob wrote in :
That's using Windows XP to manage wireless instead of the card vendor's utility. Check for an option to do that.
It's actually pretty easy to use -- install and give it a try.
Reply to
John Navas
John, I got netstumbler and installed it on the laptop which has a WPC54G talking nicely to a WAP54G. But after about 30 minutes of trying everything I could see, it continued to say "no wireless adapter found" (or close to that). I found at least one place that indicated that the Linksys PC Card adaptor is supported but I never found "WPC54G" explicitly given. So, if I could get past this one problem, perhaps I would find it easy to use.
I never saw it mentioned but I presumed that the Laptop with the "client" adapter (I think it is called) is the place to use netstumbler. Just in case I put it on the machine wired to the WAP54G and got the same response - no adaptor found.
I'll play with it some more and maybe I'll "stumble" into how to make it work. Thanks.
John Navas wrote:
Reply to
rob
Check the available adapters under Device in Network Stumbler -- I have to specify the NDIS driver to get it to work with my Atheros-based wireless adapter.
On Mon, 05 Mar 2007 19:25:34 -0600, rob wrote in :
Reply to
John Navas
rob hath wroth:
Find the check box under the properties for the wireless device with something like "Let Windoze Manage the Connection". With it checked, you get Windoze Wireless Zero Config. Unchecked and it goes to the vendor supplied client. If you uncheck the box, you may need to reboot as the Windoze Wireless Zero Config service is not automatically killed until the next reboot. If you're in a hurry, just go to: Control Panel -> Admin Tools -> Services and find the line for Windoze Wireless Zero Config. Hit "Stop".
Install and operation is quite simple. If your wireless uses an NDIS 5.1 driver (most do), then it should work.
Borrow one. I the opposite problem.... too many laptops. It's handy for testing, but keeping them up to date is a time burner. You're probably better off with just one laptop.
Ask and ye shall receive. The minimum Eb/No is the minimum signal to noise ratio that 802.11 will operate at a given speed. Note that it varies radically with speed.
Speed Min Modulation Typical Noise Eb/No Sensit Floor mb/sec dB dBm dBm 11 7.0 CCK -82 -89 5.5 6.0 CCK -85 -91 2 1.6 DQPSK -86 -87.6 1 -3.0 DBPSK -89 -86 54 24.6 64QAM/OFDM -71 -95.6 48 24.1 64QAM/OFDM -71 -95 36 18.8 16QAM/OFDM -78 -96.8 24 17.0 16QAM/OFDM -79 -96 18 10.8 QPSK/OFDM -82 -92.8 12 9.0 QPSK/OFDM -84 -93 9 7.8 BPSK/OFDM -87 -94.8 6 6.0 BPSK/OFDM -88 -94
What happens is that the your router does its best to maintain a constant BER (bit error rate). If it climbs, the router slows down the wireless connection speed until the BER improves. Actually, it uses PER (packet error rate) which is easier to calculate but the effect is the same. What the user sees is a constant S/N ratio which is directly proportional to the error rate.
If you want to see the SNR change, dive into the routers wireless setup and change the connection speed from "auto" to 54Mbits/sec. That will yield a radically varying SNR with signal strength.
Yep, that's the way it works for everything else remaining constant. In this case, the varying connection rate and modulation method, and the routers attempt to maintain a constant BER/PER, is what's causing an almost constant SNR. Even so, the noise level should NOT be higher than the signal level, so something odd is happening.
Reply to
Jeff Liebermann
Jeff, thanks for all the info. I'll definitely try again with the netstumbler and look for a way to make it use NDIS.
On those dB figures, I'm using Linksys WAP54G/WPC54G, which type of modulation do you figure it's using?
Jeff Liebermann wrote:
Reply to
rob
Jeff, I just tried netstumbler again. In the device menu the only option is "use any suitable device". One line above that is an Intel something but it is greyed out. Also I did find a site that said my specific Linksys WPC54G card is supported. So, while that very card is nicely linked to my AP, netstubler just won't do anything at all. The only version available that I can find is 0.4. Doesn't sound like it's been around too long.
I looked around and found a Kismet but it must be for Macs. I also found and tried a UfaSoft product freebie. Damn, it finds the IP's of all my wireless components but there's no way I can get it to sniff the area. And as expected for a freebie, there's no documentation that really tells anything. I'll keep looking around; this sniffing sounds like a neat feature and maybe eventually I'll get it.
Thanks again, Rob
Jeff Liebermann wrote:
Reply to
rob
I don't think you have to search for NDIS. Just install an run Netstumbler 0.40. If it works, you have an NDIS driver. Just about everything for XP and w2K use NDIS drivers.
The modulation method varies with the connection speed. Get the operating wireless speed from your WPC54G connection manager, and just look up the corresponding details in the chart supplied.
Reply to
Jeff Liebermann
That could be your ethernet. Your WPC54G should be recognized, which apparently it's not. No clue why it's failing.
That's the current version of Netstumbler. Why update something that works just fine?
There is a MacIntosh version, but Kismet is primarly written for Linux.
Well, if you're shopping for a connection manager, try:
Works fine for me, with one gotcha. It installs a WiFiHopper service which seems to interfere with other Wi-Fi sniffers. Try it, but be prepared to disable the service if you try others: Control Panel -> Admin Tools -> Services
Reply to
Jeff Liebermann
I tried it and amazingly it does see my AP. But it's a freebie and not much else works.
The signal level it shows stays at -64dBm(/Mhz I presume) even though I cover my antenna with my hand. My Linksys measurement shows signal dropping but hopper level never changes. Clearly it isn't finding the actual level.
It doesn't find my neighbor's AP, no filters on. Linksys monitor does find it. I guess my Linksys monitor is a form of sniffer but all it does is list the ssid and strength, no other info is given. But it gives more than hopper.
Apparently the hopper didn't interfere with the Linksys monitor, it's still working the same.
Also the HELP menu item doesn't work so I can't really study how to use the thing. Maybe if I pay for it they will give me the help files.
So I still don't have anything that does much sniffing but I'll keep looking. Thanks for your input. Rob
Reply to
rob
rob hath wroth:
It? Oh, WiFiHopper. It's not exactly free. The sniffer part of the puzzle is essentially free. The connection manager part times out after 15 or 30 days. The connection manager part is actually quite handy (for me) as it will allow me to connect by MAC address instead of SSID. That's useful when I have to deal with a network of access points, all with the same SSID.
Use the graph, not the table to get your numbers. The table does not update in real time.
Are they broadcasting their SSID?
The Linksys monitor is nifty. It shows access points that don't broadcast their SSID.
Nope. It's an active monitor and uses probe requests just like Netstumbler and WiFiHopper.
Actually, it does interfere. When you have WiFiHopper running, it shuts down any other connection manager including Windoze Wireless Zero Config, and takes over using its own connection manager. Please read the docs. This is fine except that after the connection manager of the program expires after a month, it still kills the Windoze WZC or any other connection manager when running. There are other interactions that bug me. Every time it probes a live connection will stall. The beta version had it probeing every 3 seconds which resulted in a download practically stopping. Even set to the current default (10 sec???) it will still slow down downloads. In short, you can't really have it probeing for access points while trying to do useful web surfing or downloading. Also, when exiting the program, it will drop the connection. That's fine, but some drivers don't recover very gracefully. I find myself rebooting a bit too often. Still, it's a useful program if you don't try to do everything at once.
Help works for me. However, I'm still running some old beta version. I'll try a later version when my laptop isn't busy.
Well, another possibility is to use Kismet running on a Linux LiveCD. Nothing to install, just boot the CD. See:
707MBytes to download. Ugh.
Reply to
Jeff Liebermann
The graph doesn't work either. It does bump along updating "hits" on horiz axis but no graph line is seen. I presumed it was just plotting constant -64dBm, which is on the lower end of the vert axis. In any case, nothing interesting is seen on the graph.
If SSID is something like "my Linksys", then yes, it is being broadcast by my neighbor.
I just found some info on the wifihopper site, maybe that will help me figure out how to use it -- since the program's HELP doesn't exist.
I'm beginning to think something is weird with my laptop system since I can't get netstumbler to do anything, and a lot of others are able to use it OK. But at least it does connect with the AP and hasn't broken down yet.
Thanks again.
Jeff Liebermann wrote:
Reply to
rob
On Thu, 08 Mar 2007 12:46:56 -0600, rob wrote in :
You do have a _unique_ SSID set on your own network, right? That's very important.
Reply to
John Navas
Yes sir. Out of the box I had the default but now it's quite unique -- after reading postings on this ng. Thanks.
John Navas wrote:
Reply to
rob
If you are interested they have just released BT2 Final. The auto-configure for Kismet works for my WG511T(Atheros) and Ralink2500 cards, in fact if both cards are fitted you get a choice of which one to use. They have included a new sniffer called Wicrawl ,which I haven't got to work with my cards yet.
formatting link
Reply to
kev
Having had a further play I still can't get it to work with my Ralink card, however I have got it to work with the WG511T.
1.Open console. 2.Type airmon-ng start wifi0 press enter. 3.This will report an ath in monitor mode, call it ath1 (due to playing I was at ath3 for monitor mode). 4.Type ifconfig ath1 up press enter (It is important that you use the one designated in monitor mode) 5.Start wicrawl ,go to Interfaces and make sure ath1 monitor mode is ticked.
There seems to be a bug in the copy I have as the Channel reporting was incorrect.
Reply to
kev

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.