nstreme and interference

rgenovesi · Wed Jul 27, 2005 12:52 am

Hi All,

Does the nstreme protocol provide any better results against interference? I thought someone said that it did (due to different packet headers?), but I searched through the forum and didn't find anything.

We are running up against 2.4 Ghz phones more and more these days -the kinds that have a base station and multiple handsets and therefor are broadcasting all the time...

Thanks,

-rob

mp3turbo2 · Wed Jul 27, 2005 11:23 am

> Does the nstreme protocol provide any better results against interference?

no, it does not. Nstreme is more vulnerable to interference than anything else I know...

tully · Wed Jul 27, 2005 12:57 pm

That is not correct. Nstreme 'WAS' having long ping times (latency) in previous versions (before rc2? or so). The throughput under our test was the same or better than 802.11. The TCP throughput (only one connection) was not good because TCP automatically slows each TCP connection based on latency. If you had multiple connections, you would get the equivalent throughput as 802.11.

Since rc2? or before, we have added a dynamic ajustment that does not increase latency when there are low quality links.

John

mp3turbo2 · Wed Jul 27, 2005 8:57 pm

to be objective, I have to say that we tried this with 2.8.27 so John's statement could be correct. But beware, John is talking about latency and throughput, which are not directly connected to resistance against interference.

Our test was performed this way : 2.4GHz, rooftop with 4 directional antennas and one sector, all using B mode. We intentionally installed another antenna pointing at the same direction as one of the strongest directional antennas there, we basically installed parallel link with existing one separated by 1m distance. Then we used the same channel like the first link had. Using nstreme (exact-framing) we could transfer something around 60kB/s with pings around 900ms (and falling out) on G-mode... switching off nstreme brought us to to 250kB/s and ~100ms stable pings still on G. We didn't test B mode.

Also, we didn't try any 2.9rc versions, so this could be outdated test. It was some week ago.

tully · Thu Jul 28, 2005 7:44 pm

You should use the dynamic framing. Exact framing will be pretty much like 802.11 in some cases.

v2.9 is different in many ways, so you should do your test with this.

John

mp3turbo2 · Fri Jul 29, 2005 10:10 am

why is this not written in manual? :)

btw, we had better experience with exact framing...

Fri Jul 29, 2005 10:23 am

is it not written in the manual? what is this:
http://www.mikrotik.com/docs/ros/2.9/in ... 7470083969

actually there is a mistake, we will change "best performance" to be for `dynamic size`

surfnet · Sun Jul 31, 2005 7:03 pm

Thanks for the claification. I had read the 2.8 manual and had my entire network on exact size

But... after changing to dynamic size everything improved

I am having an issue where evey morning my signal swings from 75 to 90 and then back again. this was causing the signal to drop and come back again. After changing to dynamic size , it does not lose its connections.. hurray

but the big question still remains. why does signal change? I can see noise going up and down, but without a noise measurement we cant tell.

Signal should never change unless you move the antenna, have a bad radio, or have a tree grow overnight..none of these things occured.. I wonder if Mikrotik is somehow combining signal and noise to come up with signal.. perhaps someone from Mikroitk could shed some light on this matter..

dont get me wrong I LOVE mikrotik, I had a mesh network with 18 nodes and it sucked!!!! all locustworld did was tell me my hardware was bad, but after simply installing Mikrotik on the same hardware, my problems went away and my service improved 100fold

but I still have that nagging problem of noise.. I am simply blind to wahtever noise is out there in the MK world.. so this is another plea....

Mikrotik gods.. please help me see noise. or at least shed some light on why you dont.

thanks!!!!

john2 · Sun Jul 31, 2005 9:57 pm

Atheros does not give any noise figures.

The reason the signal strength changes is because the Atheros has different transmit powers for each rate. The rate changes depending on the quality of the link and therefore, if the rate is slow, the signal strength is higher.

John

wildbill442 · Sun Jul 31, 2005 10:26 pm

Thanks for the claification. I had read the 2.8 manual and had my entire network on exact size

But... after changing to dynamic size everything improved

I am having an issue where evey morning my signal swings from 75 to 90 and then back again. this was causing the signal to drop and come back again. After changing to dynamic size , it does not lose its connections.. hurray

but the big question still remains. why does signal change? I can see noise going up and down, but without a noise measurement we cant tell.

Signal should never change unless you move the antenna, have a bad radio, or have a tree grow overnight..none of these things occured.. I wonder if Mikrotik is somehow combining signal and noise to come up with signal.. perhaps someone from Mikroitk could shed some light on this matter..

dont get me wrong I LOVE mikrotik, I had a mesh network with 18 nodes and it sucked!!!! all locustworld did was tell me my hardware was bad, but after simply installing Mikrotik on the same hardware, my problems went away and my service improved 100fold

but I still have that nagging problem of noise.. I am simply blind to wahtever noise is out there in the MK world.. so this is another plea....

Mikrotik gods.. please help me see noise. or at least shed some light on why you dont.

thanks!!!!

noise is not a mikrotik problem, noise is part of radio frequency. You will always have some level of noise in any system you use regaurdless of the manufacturer.

As john noted, the signal changes depending on what rate you're connecting with.. the rates will fluctuate due to a fluctuating signal and uses up valuable resources to reestablish a connection.. Only use the rates you need. If you're not pushing 24+mbps through the link then you don't need to associate @ 24mpbs... so unselect the supported rates that aren't being used. This will prevent your link from renegotiating link speed when the signal fluctuates, and will use less resources to maintain the link.

surfnet · Mon Aug 01, 2005 12:18 am

My APs are all locked at 24 or lower connection rate and the signal on 2 out of 4 of the APs connected it to it show this swing.

I have locked the rate to 6 and it still swings wildly

quote
Atheros does not give any noise figures
end quote

that is a bit disturbing.. does anyone know why Atheros would do that?

would it be safe to assume that the chipset is reporting both signal and noise in the same figure.. so that more noise would equal less signal ?

mp3turbo2 · Mon Aug 01, 2005 1:19 am

signal won't change from -75 to -90 because of datarate change.

oh btw, we tested nstreme with latest rc7. Still my results from 2.8test are supported by latest finding - have to say that this is the least resistant thing (=worst) I've ever seen in my life. Standard 802.11b (802.11g) equipment is much faster in interfering area than Nstreme. Seems to me like Nstreme is high-tech toy requiring PERFECT conditions to work properly.

bye, mp3turbo.

sten · Mon Aug 01, 2005 1:40 am

I dont think the rates change by a whopping 15 dbm. I could be wrong as radio math is not my strongest and misunderstanding the specs is a possibility. Just to mention a few reasons that we've seen cause that;
Water in antenna/cables. Power fluctuations(!!!!). Short circuitting the antenna cables. Chip's that are failing (remember, most cards out there arent designed to live for more than a few years!! Also alot of it is consumer grade quality and have a very high failure rate) and bad connectors.

john2 · Mon Aug 01, 2005 9:20 am

MP3, you should send your supout file to support@mikrotik.com, most likely you have changed the configuration to make problems.

Your comments are incorrect according to ours and others testing.

stephenpatrick · Mon Aug 01, 2005 10:58 am

I'm staggered to hear that Atheros chipsets doesn't give SNR or Noise Floor readings.
I would assume any radio does that - at least one older chipsets did I believe.
I'm concerned about how to "fault find" densely located MT setups and sectored base stations, if there is no measurement of noise floor at the receiver.

There's an interesting article here, well worth a read:
http://www.connect802.com/download/tech ... 100201.pdf

Regards

surfnet · Mon Aug 01, 2005 4:59 pm

I read this on another forum.. perhaps it is a clue..

--Quote
It has been discussed that Atheros numbers are reported more like a SNR than a RSSI. The way I understand it, as noise goes up, Atheros cards report that as lower signal, but of course, that may not really be lower signal.
--end quote

mp3turbo2 · Mon Aug 01, 2005 6:05 pm

yes I can and I will, but it will happen on weekend only. We have to announce link failure to customer first and even with this, perform test at night... My observations so far : we had another radio client exactly in the same direction as the original one, so we tried to connect the new one to the same AP. We got disconnections all the time - connecting and immediately disconnecting, in endless loop. Changing frequency 10MHz up or down solved the problem, client is connected and stable. Problem : interference, there is one close channel (not identical!) used in the second client locality. That's it.

tully · Mon Aug 01, 2005 6:11 pm

Also, please send a supout file of the working system -- so that we can compare any changes.

John

tlkhorses · Tue Aug 02, 2005 3:38 pm

Maybe I am missing it but I do not see noise reported anywhere for prism cards in routeros either. Missing feature?

tk

surfnet · Tue Aug 02, 2005 4:35 pm

well now we know why the Atheros shows no noise. I have a feeling the Mikrotik doesnt really care to much about Prism and therefore insnt going to take much time in developing a noise setting that only works on those cards.. it would be a GREAT feature that would probably help out alot.

nstreme and interference

nstreme and interference

Who is online