Thoughts on Wide Channels

[sfx2000]

Just wanted to share - pulled out an old-school 802.11n AP for some testing - I like this one, as when turning on 40MHz channels, it does it, doesn't really care, just does... Linksys WRT160NL with 1.0.04 build 2 firmware...

And I have an old-school 802.11n client that pretty much does the same - if the AP is Wide, it'll do it...

And did some testing... looks kinda scary a bit, if you're a neighbor - but then also consider - AP is on CH1, the secondary is on CH5, and 6 ft away, this is 2 meters for you metric folks, the secondary channel is down in the noise floor - below -70 dB RSSI - which means for the neighbor next-door/across the street, might not be much of a problem - might be more for CH1 than for CH5, just because...

This is an H-264 high-bit rate session across the house, about 65 feet (20 meters) from the AP to the client...

Someone across the street on CH6 probably won't see it at all... they'll see CH1, the primary, they won't see the secondary, and they'll likely stay on CH6 because it's the right one to choose for them...

CH11 - that's my primary Ac1900-class AP's - their sitting idle for the moment, as they're working hard on the 5GHz. I'll probably move them over to CH1 to free up 11... or not...

Remember beacons are hot, they have to be - but when you look at traffic...

 

[sfx2000]

And here's why it looks scary.... but it really isn't... next post, if discussion permits, I'll say why...

 

[dvohwinkel]

And here's why it looks scary.... but it really isn't... next post, if discussion permits, I'll say why...

View attachment 6126

tease

 

[stevech]

Those WiFi analyzers are not very useful unless they dwell and measure percent channel utilization. That is, how busy each channel is. Of course, this is time of day dependent and varies with users' activities. A 2 hour streaming 1080P video on WiFi at 40MHz is not so nice. But fleeting.

 

[sfx2000]

Those WiFi analyzers are not very useful unless they dwell and measure percent channel utilization. That is, how busy each channel is. Of course, this is time of day dependent and varies with users' activities. A 2 hour streaming 1080P video on WiFi at 40MHz is not so nice. But fleeting.

@stevech - the first shot is a spectrum analyzer - not a WiFi discovery tool

 

[Nullity]

@stevech - the first shot is a spectrum analyzer - not a WiFi discovery tool

I have dreamt of owning such a thing.

What are you using (assuming it's not DARPA/NSA supersekrit)?

 

[sfx2000]

I have dreamt of owning such a thing.

What are you using (assuming it's not DARPA/NSA supersekrit)?

Old Metageek WiSpy...

 

[stevech]

I"ve not seen a free WiFi analyzer that measures channel utilization (busyness). The focus on signal strength which is irrelevant for a survey - what you need to know is that on a long term basis, which channels have the least use (utilization). Most folks don't get this. WiFi is like ye ole party line phones. Ideally, you don't have to share very much talk-time.

 

[Nullity]

I"ve not seen a free WiFi analyzer that measures channel utilization (busyness). The focus on signal strength which is irrelevant for a survey - what you need to know is that on a long term basis, which channels have the least use (utilization). Most folks don't get this. WiFi is like ye ole party line phones. Ideally, you don't have to share very much talk-time.

Kismet.

 

[stevech]

Kismet... new to me. Thanks.

Check out Paragraph 3, Quick start...

https://www.kismetwireless.net/documentation.shtml

 

[sfx2000]

I"ve not seen a free WiFi analyzer that measures channel utilization (busyness). The focus on signal strength which is irrelevant for a survey - what you need to know is that on a long term basis, which channels have the least use (utilization). Most folks don't get this. WiFi is like ye ole party line phones. Ideally, you don't have to share very much talk-time.

Kismet can do it in real time - and one doesn't have to scan all channels, can lock it on to a single channel...

Wireshark, some can do this in real time (depends on OS/Drivers), but can definitely do it after the fact using the protocol statistics tool inside...

 

[Razor512]

See if you can get both of your APs on 40MHz channel width, and then have a client on each download a large file from a NAS. or have each one do a 24 hour stress test.

I also wonder, will the spectrum use look any different if a 4 stream router is in use?

They should make the routers offer a mode where they can use every available channel at the same time. Getting even wider channels (especially if they can get a 1 watt transmit power while at it), could allow for a decent balance of coverage, and throughput on the 2.4GHz band.

 

[stevech]

1 watt on the router/AP WiFi creates a very unbalanced bi-directional link since the client devices are typically 0.03 to 0.06 watt transmitter power.
The access device grossly out-talks the clients (in terms of range).

Like a rock band's PA system. Or a cop with a 5W walkie-talkie and the moutaintop repeater is 150W.

 

[sfx2000]

See if you can get both of your APs on 40MHz channel width, and then have a client on each download a large file from a NAS. or have each one do a 24 hour stress test.

I also wonder, will the spectrum use look any different if a 4 stream router is in use?

They should make the routers offer a mode where they can use every available channel at the same time. Getting even wider channels (especially if they can get a 1 watt transmit power while at it), could allow for a decent balance of coverage, and throughput on the 2.4GHz band.

4 stream is going to look the same as any other in 2.4GHz - the primary channel will always be higher, as it contains the beacon, and management messages - Beacon is always transmitted as DSSS, not OFDM, and management messages, depending on the AP mode, may be transmitted in DSS, but most modern AP's will transmit 24Mbit using OFDM..

Some asked what would happen if we move the other two AP's over to CH6, where the WideAP is overlapping - pretty interesting, the 20MHz AP's actually aren't affected much - as the WideAP has to check to see if the channels are clear...

in 11n, this is done via CTS/RTS handshaking, and as such, it needs to check both the primary and secondary - and this is a big deal, as if CTS doesn't pass on both channels, the WideAP doesn't transmit - and this goes for both stations - the AP and the client.

11ac does this a bit different, as it has Enhanced CTS/RTS, and if the secondary is blocked for whatever reason, the primary, if the handshake for that frame intent passes, it'll transmit at least on the primary...

See below;

The take away here - in a congested environment - Wide Channels suffer much more than 20MHz channels - so my recommendation is not to run 40MHz channels at all - which goes against conventional thinking that WideChannels interfere with adjacent networks - this is not the case...

There's another reason, but I'll get to that in another post on this thread...

 

[sfx2000]

Just wanted to revisit this thread...

For the last couple of months I've been running a WRT1900acV2 on Wide Channels - mostly just to accomodate a remote that is a 3R2T wide client bridge...

Primary on 11, secondary on 7...

 

[thiggins]

Just wanted to revisit this thread...

For the last couple of months I've been running a WRT1900acV2 on Wide Channels - mostly just to accomodate a remote that is a 3R2T wide client bridge...

And so what are the new insights you want to share?

 

[sfx2000]

And so what are the new insights you want to share?

Sure - see the trace below...

I've been doing some testing with 20 and 20/40 in the ISM band (2.4GHz) - what I've found...

1) Beacons and the Primary are going to be fairly 'active' - that's because the Beacon is transmitted at the basic rate, so it's going to be much higher relative gain to the traffic
2) Overall channel utilization is much lower than what folks generally consider - even stressing things with multiple streams, I hit about 25 percent of overall load on a single narrow channel
3) Airtime - this is the big one...

With Airtime - bandwidth not used is bandwidth wasted - given that if I have 100MB to move across the link, a wide channel is going to give X - when we cut the channel in half - we still get X, but it takes twice as much time on the channel...

And time is the fire in which WiFi burns...

In the spectrum above, look at the secondary channel, it's below the threshold for active interference to adjacent networks... and with 11n/ac - co-channel and adjacent channel interference rejection limits are much lower than what we saw with 11b/11g legacy...

Newer 11n/11ac chipsets also improve on rate determination - old-school used to slow down rates, but I've seen more recently that they maintain rates and rather retransmit if needed - in other words - stay fast and burn through it - wireless has always been a noisy and burst filled medium.

Also note within 11n - the AP has to check the secondary channel before it can transmit on it - if the Clear Channel Assessment says that the channel is not good, it won't - so one gets at least 20MHz, if the channel is clear on the primary.

Goes without saying - if I have two AP's 2 meters from each other, they're going to fight, and both will lose, but generally, you won't find this in the real world, and because of distance, generally we have a 10LogR function...

 

[thiggins]

So TLDR; is go ahead and run 40 MHz channels in 2.4 GHz?

 

[sfx2000]

So TLDR; is go ahead and run 40 MHz channels in 2.4 GHz?

I would say for some cases, it might worth trying - even when set to 40MHz, the primary is still 20MHz, and the secondary is conditional and opportunity driven... and there, try to park the primary on CH1 or CH11

Would be similar with 11ac and the 80+80 and 160MHz schemes... the newer 11ac based chipsets are that much better than the legacy b/g and early 802.11n...

 

[RMerlin]

The last time I used 40 MHz channel width on the 2.4 GHz band, the customer's Internet would cut off whenever his cordless phone rang (this was before the DECT 6.0 days).

Most devices will downgrade to 20 MHz whenever there's any interference. Between microwave ovens, Bluetooth and neighbours, this will most likely happen a lot.