What's new

Many Devices Ignoring 40 MHz Coexistence - Wi-Fi Certified

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

Darknessrise

Occasional Visitor
Even with the recent hard push of Wi-Fi Certified 40 MHz coexistence mechanisms, many routers I have seen pretty much ignore everything and bond channels no matter what. Some have certain settings you must choose to make it work. Why are these products slipping by and are still getting certified for "40 MHz operation in 2.4 GHz, with coexistence mechanism" is what I wonder.

First off, my EA6300V1 is certified under that. Sure, it won't bond when I set it to auto width mode while in mixed mode. On the other hand, if I set it to auto 20/40 and N only, it'll bond no matter what's going on. - Wi-Fi Certified

Secondly, the RT-N66 is certified, yet it has 40 MHz only mode. How does this get by? - Wi-Fi Certified

Some other examples(my neighborhood has a lot of networks, there's no way a 40 MHz would be allowed, especially with all of the overlapping networks):
- My old R6250 would force bonding when the "Enable 20/40 MHz coexistence" box is unchecked. Pretty much allowing the device to have a 40 MHz only mode. - Wi-Fi Certified
- My neighbor's rented Comcast gateway(Arris TG862G) pretty much always channel bonds on channel 1 even if I plop my network on theirs(while I have a strong signal) and won't let go of that bond no matter what I try. - Wi-Fi Certified
- A friend's rented Comcast gateway(SMCD3GNV) pretty much bonds no matter what. They could have another router right on the same channel/overlapping it and it won't do anything about it all. When I had that gateway, it would bond all of the time too even if I was already overlapping my neighbors. - Wi-Fi Certified
- My other neighbor's WNR1000 bonds channel 2+6 even if I'm on 1 or 6 and their gateway/printer's network is already on 1. It even went to channel 6(my channel) and kept bonding along with 6+10. - Wi-Fi Certified
- Here is a sort of situation. My WD N750 would sometimes slip up and allow bonding until I connected with my Netgear WNDA4100(good), but otherwise wouldn't bother to stop the bonding with other devices and all of the overlapping networks. - Wi-Fi Certified
- Recently was helping another friend with their Belkin F9K1116. As a test, I put it on a channel with a lot of other networks around it. Nothing really special was needed other than a basic switch on of 20/40 MHz from 20 MHz and it begin bonding no problem. The Belkin even had a G device connected. - Wi-Fi Certified


Am I missing something? Is the Wi-Fi certified somehow only based on defaults and it doesn't matter once the settings are changed? If so, I think that's silly of a thing to do.
 
Last edited:
My understanding is that it is based on the defaults for the wireless standard (B,G,N etc).
The 5Ghz bands seem to be utilized better but the 2.4Ghz band is a mess. One thing I really do not understand. If you have a wireless N router and you are using either N mode only or mixed mode G/N, I think you should only have the option to pick either channels 1,6 or 11. No others. I can't understand why they let people pick overlapping channels when using wireless N in the 2.4Ghz band. Whats even worse is when someone sets their router to like channel 3 and then selects 40Mhz wide channels and it bonds channel 7 to 3. Now they are effectivly using the entire 2.4Ghz spectrum since channel 3 overlaps 1 and 6 and channel 7 overlaps 6 and 11.
 
Most chipsets that support 40MHz mode will generally default into either channels 1, 6, or 11 for the primary channel.

The challenge is to find that Primary, and set your channel so it doesn't interfere...

Wide Channels in 2.4GHz is becoming less and less of a benefit due to co-channel interference...

The big challenge at the moment with 2.4Ghz is the QAM256 proprietary extensions on 802.11n, as there is no common ground here, and it's causing interop issues with legacy 802.11 b/g/n clients, even more so when the offending AP is adjacent...
 
I agree this is a mess. The Wi-Fi Alliance will only reveal details of the certification test to members. One thing I do know is that the 802.11 spec does not specify a timeframe that 40MHz coexistence must be recognized in. Could be 30 seconds, could be 10 minutes.

The other thing I have asked NETGEAR about is the "disable 40 MHz coexistence" controls they have. NETGEAR said that the Alliance originally did not allow this, but now does.

So, essentially, if you want to set your router to 40MHz mode in 2.4 GHz, in many cases you can. Or the router will even do it for you.
 
Maybe back then it used to be that to get WiFi certified, router must be on mixed mode with 40 MHz coexistence by default and not have an 40 MHz only option. If there's a 40 MHz only option (aka Up to 300 Mbps mode on some Netgear routers), the 40 MHz coexistence will still be in effect.

This is true to Netgear WNDR3400v1 with stock firmware where 40 MHz coexistence is still in effect despite selecting "up to 300 Mbps" mode. Thus the user cannot really disable the 40 MHz coexistence in this case without using a 3rd party firmware.

In my WNR3500v1 with stock Netgear firmware, selecting the 300 Mbps mode results to forced channel bond. In this case, selecting that mode actually disables the 40 MHz coexistence.

Nowadays, (this is just my thought) the WiFi certication requirement is to have the router be in 40 MHz coexistence out of the box and as a default while having the 40 MHz forced bond as a customized option. That's probably how those routers/AP mentioned by OP got certified.

As for "true" WiFi certified router, the Linksys WRT1900AC surely deserves the certification IMO due to not having a "40 MHz only" mode on both 2.4 and 5 Ghz frequencies so the user cannot disable this option without using openwrt.
 
At least in terms of choosing a non-primary channel, there are reasons. There can be a non-wifi interferer, path optimizations, client issues, etc.

For instance, one of my clients, if I have channel bonding setup on channel 1+5, it is slow as molasses. If I set my AP to 2+6, it works great. Which is critical, because my outside AP has issues if I set it to channel 1 or 2. Works great on 3+. So if I want non-overlapping and things to work great (based on my OLD router/AP), I had to set it to 2+6 and my outdoor and basement APs to 11. Non-overlapping because of the way channel bonding works. Also the way the AP/router setup channels, I couldn't set it to channel 6- for the channel bonding. It would only do primary channel + (IE set channel 1 and it'll do 1+5, set channel 6 and it'll do 6+10) with the exception of setting it to a high channel, like 11, it would set it as 7+11.

No issues now as my current WDR3600 don't have that issue doing 1+5 with my client(s).

Of course I can be a spectrum hog as I have no near neighbors and effectively no visible 2.4GHz networks near me. So 40MHz is completely kosher (and yummy!). I just wish we had through channel 12/13 in the US so I could do 1+5 and 9+13 for two non-overlapping 40MHz 2.4GHz networks.
 
Well, that surely is stupid. The good thing is that it's at least hidden away some what.

Have you tried what I can on my Linksys EA6300V1 where you can select N only and 20/40 MHz auto mode to force bonding?

It only works when you connect to the 2.4 GHz SSID immediately (when SSID shows up on Windows) right after applying the change to the the 2.4 GHz settings on WRT1900AC. However, once you trigger WiFi scan as I did by clicking the WiFi icon on Windows after the established connection, the channel bond disappeared :( and I reverted back to 20 MHz mode. Thus you cannot disable 40 MHz coexistence with WRT1900AC stock firmware on 2.4 GHz (True WiFi Certified).

Fortunately, I captured a screenshot where 40 MHz channel bond was disabled (per Windows "netsh wlan show interfaces") but my link rate still shows 40 MHz width. I clicked the WiFi icon after this screenshot then my link speed went from 400 Mbps to 173 Mbps (TurboQAM)

Capture.jpg
 
Google Fiber rolled out last week in my apartment complex. And boy, is 2.4GHz a mess now. The router that they supply (which is very difficult for most people to replace with their own because it requires both VLAN and CoS tagging--doable if you have DD-WRT/OpenWRT/Tomato, but impossible otherwise with consumer hardware) defaults to 40MHz on 2.4GHz regardless of what the interference situation looks like.

The last time I checked, there were half a dozen Google Fiber networks, of approximately equal strength, literally right on top of each other, all blaring away at 40MHz (four at 4+8, and one each at 3+7 and 5+9). 2.4GHz was always pretty bad here, but now it's virtually unusable. Their box does do 5GHz, but you know how few client devices on the market support 5GHz. I had swapped the WiFi cards in all my laptops with Intel 6xxx ones a while back, so I never use 2.4, but man, I feel sorry for my neighbors.

And by default, there is virtually no configuration for it: the basic settings allow you to enable/disable WiFi, set the SSID, and set the key. That's it. You have to check a box to allow advanced settings (which also reboots the device) before you can change anything else.
 
Last edited:
Google should certainly know better... Who makes the Google Fiber box?
 
With the blazing speeds of Google fiber is it even possible to obtain full speeds with only 20 Mhz channel width ?
 
With the blazing speeds of Google fiber is it even possible to obtain full speeds with only 20 Mhz channel width ?

Seeing as how full speed is gigabit, no, you won't get full speeds even with 11ac with a 80Mhz channel width on 5GHz. (Their device is 3x3 dual-band 11n, so it has a nominal speed of 450 per band.)

The problem isn't with 40MHz on 2.4GHz. It's 40MHz on 2.4GHz when 2.4GHz is blanketed by a dozen other networks. With the kind of congestion you find in an apartment complex where you have neighbors next door, upstairs, and downstairs, interference is so bad that you don't actually gain anything by using 40MHz and you only make the problem worse for everyone else.
 
Seeing as how full speed is gigabit, no, you won't get full speeds even with 11ac with a 80Mhz channel width on 5GHz. (Their device is 3x3 dual-band 11n, so it has a nominal speed of 450 per band.)

The problem isn't with 40MHz on 2.4GHz. It's 40MHz on 2.4GHz when 2.4GHz is blanketed by a dozen other networks. With the kind of congestion you find in an apartment complex where you have neighbors next door, upstairs, and downstairs, interference is so bad that you don't actually gain anything by using 40MHz and you only make the problem worse for everyone else.

Agree i dont live in a apartment and 2.4 Ghz is so crowded here its almost unusable so i disabled it and stay on 5 Ghz only works well and believe it or not the range with the 68U is excellent almost as good as 2.4.
 
How does 5 GHz look like with those Google Fiber routers?

So far, I've been really lucky that nobody's stepped on my toes (yet). Before Fiber, there were only three 5GHz networks that I can see from my home: mine at 44-48, another at 36-40, and another at 149-153. The six new Fiber nets are one at 36-40, three at 149-153, and two at 157-161.

5GHz is gonna be so much fun in several years when 80MHz becomes more common.
 
So far, I've been really lucky that nobody's stepped on my toes (yet). Before Fiber, there were only three 5GHz networks that I can see from my home: mine at 44-48, another at 36-40, and another at 149-153. The six new Fiber nets are one at 36-40, three at 149-153, and two at 157-161.

5GHz is gonna be so much fun in several years when 80MHz becomes more common.

I can't argue with that. Out of curiosity though, how close are your homes/construction? Townhomes?

I only ask because 5GHz really just doesn't go that far. When I was living in a townhouse, 2.4GHz was lousy, but I couldn't pick up a single 5GHz network. I'll grant this was 2 years ago and also with my community I knew of exactly two "techies" who probably had something other than the comcast/verizon supplied all-in-one (IE 2.4GHz only). Both were about 5-8 houses away.

When I stopped by my house a week or so ago the only 5GHz I saw was my renters network, though that doesn't mean the other 5GHz networks around were too weak, could just mean no one else had any 5GHz networks nearby.

My current house that I live in, I can barely see any 2.4GHz networks, let alone 5GHz. My own 5GHz if I step out of my house I can't pick up the networks more than 50ft from my house.

I won't disagree with concern on 5GHz 80MHz. It takes up a lot of spectrum and without DFS, you really only have two non-overlapping ranges. My bigger concern is with 160MHz. There is exactly ONE network that can occupy that without DFS.

You've got 180MHz of range in 5GHz without DFS, channel 36-48 (80MHz) and 147-165 (100MHz). I haven't seen a lot of routers that even allow channel 165 (even though it is allowed in the US). Very few that enable DFS.

If you did throw in DFS as it now exists (post June 1st FCC changes), you have, IIRC another 240MHz of spectrum. Granted, some of it is VERY low power, just 50mW, but a bunch of it is 250mW. I just hope that the FCC decides in the end to "release" the 80-100MHz chunk that is under discussions for vehicle peer to peer networking to unlicensed non-DFS space and make it full 1W. That would at least boost it to close to 300MHz of non-DFS full power unlicensed spectrum. That and then make DFS mandatory for 802.11ac revision 2015 or something compliance.

Honestly, considering the "what up" with spectrum and penetration, I think the FCC would be better off allowing the 3.6GHz chunk for vehicle peer-to-peer networking instead.
 
I think some routers default to that in order to improve wifi performance. Others will default to it on but can sometimes offer the option to disable it.

Think of it as the uncheck for awesomeness option

While it can potentially be an issue if you have a ton of routers in the same area all wanting to use 40MHz, but usually when it is disabled, 2.4GHz throughput does increase by a good amount.

e.g., here it is on one of my routers.The only way to get 40MHz while in the city is to uncheck that option, if not, you will likely never be able to use the full performance of your router.

http://i.imgur.com/ES3rYZx.jpg


The 40MHz mode should be able to detect when the 2.4GHz band is already screwed and just understand that the 40MHz mode will just be a drop in the ocean and thus should just enable it anyway.

for example, So far there are around 160 or so wifii networks in range on the 2.4GHz band all of the channels are pretty crowded and thus having a router drop from 40MHz to 20MHz, will not really help anyone, but will lead to slower wifi performance for the owner of the router.

5GHz is still relatively clean with while indoors, the only 5GHz network that is not my own is some cisco access point owned by timewarner. Other than that, I can only pick up other residential 5GHz access points if I go outside, or use my yagi antenna while indoors.
 
Last edited:
Google Fiber rolled out last week in my apartment complex. And boy, is 2.4GHz a mess now. The router that they supply (which is very difficult for most people to replace with their own because it requires both VLAN and CoS tagging--doable if you have DD-WRT/OpenWRT/Tomato, but impossible otherwise with consumer hardware) defaults to 40MHz on 2.4GHz regardless of what the interference situation looks like.

Oh wow...

Good news here perhaps - most tools like InSSIDer and similar look only at the Beacon Frame - and the information elements contained within - so an auto 20/40 mode usually shows up as 40MHz - depends on the driver otherwise...

Why is this important to note - 802.11 b/g/n/TurboQAM - in 2.4GHz, the beacon frame is always transmitted according to 802.11b specs - so it's lower order modulation and very high coding gain - this is so that all 802.11 capable devices can see the AP transmitting the beacon - within a BSSID set, this is also important for the attached clients/repeaters to sync to the main/primary AP.

In 802.11b/g - Co-Channel/Adjacent Channel rejection minimum performance requirement was 20dB - meaning that for a given client, you needed to see signal 20dB higher from the primary AP as opposed to a co-channel/adjacent channel AP, either as part of the common SSID set or not...

With 802.11n - this requirement was changed to -9dB - anticipating that WiFi would be more popular (along with other technologies in the ISM band) - so while you might see a lot of AP's transmitting Beacon Frames (see above), the real impact perhaps is not as bad as it sounds...

Goes without saying though - Beacon Frames do take up time on the shared access channel, and they contribute additional noise, but in 802.11n space it's not as bad as it appears to be.

sfx

(note - all bets are off with the QAM256 approaches, as these are clearly non-standard in the 2.4GHz band - 11ac is 5Ghz only, and vendors should not be doing this at the risk of being bad neighbors...)
 
Actually that is not true, on some routers you can change the beacon frame modulation so it is transmitted at something other than 1Mbps. In fact that is normally what is done if you set a router to 802.11n mode only (YMMV depending on the router firmware).

The beacon doesn't have to be transmitted at a low order modulation, its just that is what is typically done.
 
Actually that is not true, on some routers you can change the beacon frame modulation so it is transmitted at something other than 1Mbps. In fact that is normally what is done if you set a router to 802.11n mode only (YMMV depending on the router firmware).

The beacon doesn't have to be transmitted at a low order modulation, its just that is what is typically done.

There's a valid reason for the beacon frame being transmitted 1Mb/Sec - while the local WLAN may be 11n/11ac, adjacent networks might not... this is also why 802.11n Greenfield was a very bad idea, and why it was removed from 11ac...

sfx
 
What issues would it cause with neighboring networks? I am not aware of any.

I know it would cause issues with older 802.11 clients that are only able to to 11b, but AFAIK it'll work fine with at least 11n/ac clients and possibly 11a/g clients as well.

Or at least when I've played with it in the past, all of the 11a/n/ac clients I have had worked just fine. I don't have any b/g clients, nor did I try operating them in 11b/g mode to see if it caused any issues.
 

Latest threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top