Community discussions

MikroTik App
 
User avatar
egenethebest
just joined
Topic Author
Posts: 4
Joined: Tue Jun 08, 2021 4:51 pm

Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Tue Jun 08, 2021 4:52 pm

Hello everyone,

We have a site (a secondary school) where all WiFi networks are managed by us. The building should in theory have no rogue APs that introduce interference, however, other RF emitters such as microwaves, game console controllers, Bluetooth devices do exist.

I, being a young still learning network engineer wanted to follow all the best practices as detailed in online resources, as well as published books about network deployment strategies. From what I found it is generally recommended to use non-overlapping 20 MHz wide channels when deploying a network in the 2.4 GHz spectrum. Configuring APs in a manner where each AP broadcasts on a channel that is not used by any other AP nearby. I have also found a lot of resources that say that using 40 MHz wide channels is generally not recommended in the 2.4 GHz band since there are only 2 non-overlapping chunks of the available spectrum and that it would be extremely difficult to divide it in a manner where APs do not interfere with each other.

Unfortunately, our senior network engineer would like to use 20 MHz Ce channels (40 MHz effective). His logic is that we should use as much of the spectrum as possible and that the best way to get a transmitter to stop occupying the spectrum is to allow it to transmit the data at the fastest data rate possible (since a higher max speed is reachable with 40 MHz). He tells me that that I do not understand how 802.11 works and that APs operating in the same chunk of the spectrum can utilize it better if allowed to use the entire spectrum due to collision avoidance algorithms such as CSMA/CA, as opposed to dividing the spectrum into smaller chunks and having each AP use their smaller chunk.

I agree with him that his solution might "better utilize the spectrum", however, I do not see it as bringing a better quality of service to the clients utilizing this network.

Could someone please explain why he is or isn't correct? I cannot find a single online resource that recommends the use of 40 MHz wide channels in the 2.4Ghz spectrum unless you only have 1 or 2 transmitters each using their half of the available spectrum. However, since he is the senior network engineer we have to do things the way he says...
 
User avatar
rextended
Forum Guru
Forum Guru
Posts: 11967
Joined: Tue Feb 25, 2014 12:49 pm
Location: Italy
Contact:

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Tue Jun 08, 2021 6:17 pm

Your "senior network engineer" was an ice cream maker before?

In fact LTE and 5G have portions of 5Mhz at random instead of 40MHz channels ... or 80Mhz ...

Thank goodness there are people like him, who make a "good mess" then call "us" and give us the pay to fix that.
 
User avatar
egenethebest
just joined
Topic Author
Posts: 4
Joined: Tue Jun 08, 2021 4:51 pm

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Tue Jun 08, 2021 8:02 pm

Your "senior network engineer" was an ice cream maker before?
LOL! 😂😂😂😂😂😂😂😂😂

Well I do not know what other arguments I can use to convince him.
He does have a background in radio engineering, and simply tells me that I do not understand how the technology works.
When I show online guides on best practices where it clearly states that 40 MHz wide channels are not recommended for 2.4 GHz deployments he proceeds to tell me that "there are a lot of stupid guides on the internet and that I should read a technical book about it instead"

I go on and get the O'Reilley "802.11n: A Survival Guide" by Matthew S. Gast https://www.amazon.com/802-11n-Survival ... 8&me=&qid=
Find the relevant sections and tell him:
Page 62:
In the 2.4 GHz band, however, there are no good choices for where to put a 40 MHz channel. Figure 6-1 shows the difficulties of enabling a 40 MHz channel in only 83 MHz of spectrum.
6711658f-a7de-429e-8498-6bb4ab3e720f.jpeg
Page 63:
As a general rule, 40 MHz operation in 2.4 GHz is acceptable if there are no other networks in range.
Page 105:
Do not use 40 MHz channels in the 2.4 GHz band. With only 83 MHz of spectrum, you have space for just two, and that’s only if you ignore the overlap problems. Just don’t do it.
He then proceeds on telling me that this is only applicable if there are networks not controlled by us using same spectrum. Since we manage the whole building then only our networks have to use the spectrum. If we had other people's networks using the spectrum, then 40 MHz would not work well.

Please tell me if I am wrong, but I do not see how two of my APs using the same frequencies in the same area is not a bad idea, but if one of the APs is someone else's on the same frequency its a totally different thing.

He then proceeds on explaining that CSMA/CA networks can easily operate in the same frequency due to collision avoidance algorithms and that since we only have one network it does not matter if a client is waiting for one AP to finish transmission or for another.


I really like my work place and it hurts me when school teachers come to me complaining that our WiFi is shitty.
If anyone with a technical background in radio engineering could explain either why he is wrong or why he is correct please let me know.
He is not taking any arguments I find on industry best practices since he says that its all fluff and that only people who understand how the physics behind radio transmission works understand how to set this up correctly.
You do not have the required permissions to view the files attached to this post.
 
User avatar
mkx
Forum Guru
Forum Guru
Posts: 11381
Joined: Thu Mar 03, 2016 10:23 pm

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Tue Jun 08, 2021 8:12 pm

Honestly I don't see how you could possibly win this argument. He is obviously very confident about his own knowledge (so he won't take any technical arguments) and he is senior to you (so you can't force your view on him).
 
User avatar
egenethebest
just joined
Topic Author
Posts: 4
Joined: Tue Jun 08, 2021 4:51 pm

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Tue Jun 08, 2021 9:25 pm

Honestly I don't see how you could possibly win this argument. He is obviously very confident about his own knowledge (so he won't take any technical arguments) and he is senior to you (so you can't force your view on him).
Well what would you do in my place? Should I just leave it configured as is and disregard all complaints about the Wireless network working like trash?
It just feels wrong ...
 
User avatar
mkx
Forum Guru
Forum Guru
Posts: 11381
Joined: Thu Mar 03, 2016 10:23 pm

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Tue Jun 08, 2021 9:52 pm

Due to lack of any serious advice, I'll resort to sarcasm. So you have 3 options:
  1. resign from your job immediately
  2. fight with senior staffer and resign from your job a bit later
  3. leave wireless config according to senior's "law" ... and move around the premises wearing paper bag over your head while waiting for the "senior" to retire.

If you go with option 3, you can always point at senior when faced with complaints. If some discussion happens "over a beer", you should be very careful not to voice your disagreement with the senior's views too hard, you never now when the word comes around.
 
User avatar
bpwl
Forum Guru
Forum Guru
Posts: 2978
Joined: Mon Apr 08, 2019 1:16 am

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Tue Jun 08, 2021 10:51 pm

I can feel the stress. It will be very difficult to convince your senior, even if you do find enough scientific studies and peer-reviewed publications that say the opposite.

Problem is , the wifi standards have evolved, and where an ac (wifi 5) system would gracefully make optimal use of the spectrum with partial overlap, this is not yet in the older wifi protocols.
With the older protocols, as you can find in the survival guides, BOTH channels must be free, or the transmission will wait.

If you are alone in the RF spectrum, and have everything under control (also all client devices and their settings) you can deploy 40MHz with success.
- IF you are not running out of SNR or signal strength (spreading energy over 40MHz makes it less dense per MHz). Client devices are weak senders.
- IF no (client) device is sending the "40MHZ intolerant bit", what would disable the use of the 40MHz channel, and reduce the throughput
- IF interface rate (MCS encoding) is not reduced more than you gain with the wider channel
- IF there is NO wifi signal above -96dB seen by the AP (AP's do interfere 6 times further away than their useful signal. Are you SURE to be in a free area? Modern chipsets are VERY sensitive, so old documents that ignore everything below -80dBm may not apply here)

So it is never pure science, and you cannot win this discussion with theory. It all depends.
But with Mikrotik you do have a very powerful tool in your hands, to find out which strategy is now better in this particular case.
- Wireless "Registration" will tell you what is actually used. Look at the TX/RX interface rate for every client (interface speed/channel width/guard)(TX/RX CCQ)(signal strength)(SNR)
- The consumed airtime depends on "interface_rate * CCQ", to account for retransmits.
- Wait time and channel-clear overhead, beacons and other admin packets cannot be seen. The only tools there are "Snooper" and "Freq usage".
- Snooper will tell you if there are other 802.11 (or nv2) transmitters around. (Printers, airco, domotica, camera, ....). Mostly this tool is eye opening, as there is more than expected.
-"Freq Usage" will also tell you if there are non-802.11 transmitters around.

So it is not yes/no, right/wrong. It depends. But it is clearly time to move up to 5GHz !!

The old document "2.4GHz is dead" disappeared from the internet unfortunately, and its too big to attach here, but I can give the conclusion:
Klembord-2.jpg
Klembord-3.jpg
Klembord-4.jpg
You do not have the required permissions to view the files attached to this post.
 
User avatar
mkx
Forum Guru
Forum Guru
Posts: 11381
Joined: Thu Mar 03, 2016 10:23 pm

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Tue Jun 08, 2021 11:08 pm

Just in case you decide to go with option #2 from my post above ... you can argument that professional networks, consisting of multiple base stations (APs in WiFi talk) and operating using single frequency channel, use pretty complicated mechanisms to overcome inter-base-station interference:
  • examples of such technologies are 3GPP release 99 (a.k.a. UMTS), 3GPP release 8 (a.k.a. LTE) and 3GPP releases 15 and 16 (a.k.a. 5G)
  • some of these technologies use same basic radio technology as WiFi 802.11n (and beyond), namely OFDM and higher-order (e.g. 256) QAM modulation
  • all of those technologies allow TDD operation (Time-Division Duplex, where both radios use same frequency to transmit, part of time transmissions are from clients towards base station and base station transmits the rest of time) and some networks use TDD for LTE and 5G networks (use of TDD in UMTS is rare)

However:
  • all of these technologies have built in advanced techniques to fight inter-base-station interference. The interference is usually less of a problem in FDD networks (Frequency-Division Duplex where one frequency channel is used for base stations transmissions and another frequency is used for clients transmissions).
  • even in FDD base stations coordinate communications with clients and hand-over clients between them. By doing that they both minimize interference and ensure seamless roaming of clients.
  • for tight coordination of base stations these are (even in FDD) well synchronised. NTP for example is not nearly good enough, synchronisation has to be down to tens of nanoseconds (NTP is good to around milli second). It's done either by using stable GPS receivers (on each base station!) of by using IEE1588v2 (PTP). This tight coordination is, for example, necessary for inter-eNodeB carrier aggregation (eNodeB is name for LTE base station) where different base stations have to synchronize transmissions for same client terminal to a fraction of millisecond.
  • for TDD synchronisation of base stations is a must even if base stations don't coordinate between each other. The reason is this (you won't believe): the whole cluster of base stations synchronises periods of time for transmitting and for receiving. And coverage overlap between clusters of non-synchronised base stations is ideally none (i.e. they don't overlap, there's no coverage area between them).
    The big reason is that base stations transmit at much higher transmission power than client terminals (e.g. 100 watts for base station VS. 0.2 watts for terminal) and have antennae with much higher gain ... and if adjacent base station would transmit at time when base station is listening for client transmissions, transmissions of client terminals would drown in transmission of adjacent base station.
  • all of these technologies feature scheduling entity in base station. This entity takes care of scheduling uplink resources to client devices as they require it (uplink is usual name for data being transmitted by client device and received by base station), clients indicate requests for uplink transmissions via special means of communication (shared signalling channel which is the only communication where collisions can happen at all ... and that's a tiny fraction of time hence performance hit due to collision is negligible).
    In case of inter-eNodeB CA, uplink can be CA as well ... and in that case scheduler in eNodeB running serving cell coordinates uplink for other component carriers ... which means tight synchronisation between adjacent base stations is vital again.

WiFi (802.11n and beyond) doesn't have any of techniques mentioned ... or are of much lower quality.

BTW, if CSMA/CA did such a great job, I guess we'd still be using half-duplex communications over wires/fibre ...
 
User avatar
bpwl
Forum Guru
Forum Guru
Posts: 2978
Joined: Mon Apr 08, 2019 1:16 am

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Tue Jun 08, 2021 11:19 pm

Honestly I don't see how you could possibly win this argument. He is obviously very confident about his own knowledge (so he won't take any technical arguments) and he is senior to you (so you can't force your view on him).
Well what would you do in my place? Should I just leave it configured as is and disregard all complaints about the Wireless network working like trash?
It just feels wrong ...
So my advice is, don't disregard complaints, that would be wrong. But do MEASURE and UNDERSTAND what is going on, and take action from there. It could be something totally different from the 20/40MHz selection what is hurting the user experience. I do know many many things that bring a wifi network on its knees, even with 1Mbps transmission, and 20/40MHz would not even change a thing. RouterOS with its tools has an advantage here over other softwares (it also has the disadvantage that you always have to do this tuning where others will adapt on their own)

Don't let the 20/40MHz debate be the "red herring" in your search for better wifi. 20/40MHz non-optimal selection will not make a "shitty wifi experience". There is something else going on. Proof you can handle this, and maybe at the end you can gain that extra x% by showing that 20 MHz performs even better than the 40 MHz setup, if conditions are right. I don't know what your regulatory domain is, for Europe (ETSI) you would have 2 40MHz channels (1-5) and (9-13), for US (FCC) you only have 1 40MHz and 1 20 MHz with channels 1-6-11.
 
User avatar
bpwl
Forum Guru
Forum Guru
Posts: 2978
Joined: Mon Apr 08, 2019 1:16 am

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Wed Jun 09, 2021 11:29 am

He then proceeds on telling me that this is only applicable if there are networks not controlled by us using same spectrum. Since we manage the whole building then only our networks have to use the spectrum. If we had other people's networks using the spectrum, then 40 MHz would not work well.

Please tell me if I am wrong, but I do not see how two of my APs using the same frequencies in the same area is not a bad idea, but if one of the APs is someone else's on the same frequency its a totally different thing.
Well it is not about the fact that there are "networks not controlled by us", it's about the fact that the overall performance of a wifi channel is defined by the SLOWEST transmitter on that channel, if all are contending for air-time with the CSMA/CA mechanism. The CSMA/CA game is rolling dices to find out who may transmit. As statistics learn each get the same number of winning deals if the numbers on the dices are equal. Point is that once you win the game you can transmit your payload. A fast interface rate (300Mbps on 40MHz/2S) will use a very short air-time compared to that slow (MCS0, 6Mbps/20MHz/1S) transmitter. The slow transmitter will be using the RF spectrum for a long time, and as such dominate the average transmission rate more. Interface rate depends on previous transmissions successes (CCQ based on retransmits, transmits fail by interference, signal noise, distortion, etc etc). So for our 300Mbps fast connection and our 6 Mbps slow connection the average interface rate is (300*6)/(300+6) for BOTH. = 5.88 Mbps. The 40MHz didn't help much, 20MHz/2S would have given : (144*6)/(144+6)= 5,76 Mbps

Still must think if a 40 MHz channel, contending with 2 20 MHz channels is still getting the same fair share of airtime. The CSMA/CA game for the 40 MHz channel must fulfill twice the winning deal, both channels must win. (and that's only 25% of the cases, not 50%)

So the challenge is to find and eliminate the slow transmitters. (Check "Registrations" for your own network. You would need to sniff all transmissions to find others)
Elimination can be adding AP for better connection, avoiding wall transitions for better CCQ, or even just refusing those low connections (access-list signal strength limit and removing low basic rate settings (that's not removing supported rates!)). The students or teachers smartphone is the weakest transmitter. Some devices do not support 40MHz or even 2S.

I don't know your setup in school. But also know that WMM is about using different dices for different kind of traffic (lower numbers give higher priority). RouterOS always rolls the bigger numbers with WMM, until some special firewall mangle rules are added to change the priority. And even then be careful as higher priority traffic in RouterOS comes with no aggregation (A-MPDU) by default. (When you eventually win the dice game, you send only a small amount of data).
 
specials7
just joined
Posts: 1
Joined: Thu Jul 22, 2021 1:58 am

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Thu Jul 22, 2021 2:33 am

I actually agree with your senior engineer. The 20 MHz channel bandwidth makes more sense when you have no control over the Wi-Fi AP locations in your environment. It'll reduce co-channel interference if all APs use 20 MHz and pick either channel 1, 6, or 11. But if you control all AP locations, you can ensure that APs using the same 40 MHz bandwidth are not in the same area. This means reduced interference and higher rates.

I believe he's saying that even if an AP's 40 MHz bandwidth overlaps with another's (partial or complete), CSMA/CA will prevent one of them from transmitting, and hence in this two AP scenario there is no interference and only deferred transmission. Of course, this doesn't account for hidden terminals.

Moving on, I think you should check the locations where teachers complain of bad Wi-Fi and measure the RSSI and compare it to the noise floor. Do it when the school is empty to see the baseline RSSI. If the baseline is already bad (e.g., -80 dBm), then it'll only get worse when school is in session. When school starts, you can also check the number of devices connected to the APs and see if an AP is overloaded. I also think the Wi-Fi can be improved a lot by also using the 5 GHz band since it can support more users. Pretty much every phone and device these days can connect to 5 GHz. Lastly, I would check all backhaul connections from the APs including the backhaul connection to your Internet Service Provider. Make sure the backhaul speeds aren't a bottleneck.
 
User avatar
bpwl
Forum Guru
Forum Guru
Posts: 2978
Joined: Mon Apr 08, 2019 1:16 am

Re: Using 40 Mhz wide channels in a 2.4 Ghz wireless network deployment

Thu Sep 09, 2021 10:55 pm

This information, might help to understand how wide wifi channels behave.

https://www.youtube.com/watch?v=lr9VVv-RkiE

Who is online

Users browsing this forum: No registered users and 22 guests