WiFi 5GHz band and wide channels

On 5GHz WiFi there are more channels and less interference, both are important for fast wireless communications. The bandwidth can even be increased multifold by combining channels.

5GHz was introduced in 802.11a, but the radios were expensive and the band didn’t gain popularity. 802.11n was defined for both 2.4GHz and 5GHz bands, which finally launched 5GHz use. The latest 802.11ac is only defined for 5GHz but all devices still support 802.11n and most also on 2.4GHz

Channels

The 5GHz band is divided into 5MHz channels like the 2.4GHz band. Fortunately only every fourth channel (36, 40, 44…) is used which provides for de facto 20MHz channel width without the overlap problems of 2.4GHz. Most devices even cannot be tuned to the intermediate channels. The whole 5–6GHz is not available since there are some forbidden channels and some channels have special restrictions.

5GHz Channels
5GHz Channels

Originally only the four lowest channels were available in the U.S. where they are called UNII-I. Later more channels have been made available, but they have several restriction for their use in the U.S.

In Europe (or in ETSI jurisdiction) channels 36–64 are restricted for indoor use only. The maximum transmission power is 200mW (23dBm), which is greater than the 100mW (20dBm) allowed for 2.4GHz, but still doesn’t quite compensate for the 6dB attenuation due to higher frequency. In access point use the maximum transmission power is practically irrelevant, since typical user devices have less transmit power. In WiFi the connection is always bidirectional so there is no point in receiving the access point if you cannot send a reply. Common default for access points is maximum power, which means 2.4GHz signal will be received 3dBm stronger, which in turn will make most devices choose 2.4GHz signal instead of the 5GHz.

On channels 100–140 the maximum transmit power is 1W (30dBm) and the channels can be used outdoors as well. In access point use 30dBm is irrelevant, but for point-to-point connections this enables long distance links (from 10km to 50km or even more). Weather radars use channels 120–128 and access points must yield to them. At start-up the access points will listen for radar signals for 10 minutes before transmitting. On other DFS channels 52–140 this start-up delay is one minute. If the access point detects a radar signal it will switch channel automatically. Most APs will play it safe and choose a non-DFS channel 36–48 which may result in overlaps in channel use.

The upper channels 149–165 are on every fourth odd channel. In Europe they can be used according to Short Range Device (SRD) specification for transmissions up to 25mW (14dBm), but most devices don’t support these channels. For access point use the 14dBm would suffice and there are no DFS or other restrictions, but the sparse client support needs to tested before deployment.

Coverage and cell size

The wavelength of 5GHz is half of 2.4GHz, which implies higher attenuation. 2.4GHz will be received at 6dBm stronger or quad-fold when compared to 5GHz signal. A 5GHz access point will thus cover a smaller area in open space and won’t penetrate walls like 2.4GHz. Because of the stronger signal many devices will rather associate with the 2.4GHz AP. The simplest solution is to reduce the transmission power of the 2.4GHz AP by 6–7dBm.

The higher attenuation and poorer penetration can be turned into an advantage to reduce access point cell size. When the AP covers a smaller area there will be less users to compete for airtime, which translates to faster data transfer. You will need more APs to cover the area, but smaller cell size is the key to high performance WiFi.

Wide channels

802.11n introduced the concept of combining channels. Combining two 20MHz channels will yield over twice the bandwidth, since there is no need for an isolation gap between channels. In 802.11n you could combine channels on 2.4GHz as well, but there really are not enough channels available. On 5GHz combining channels is actually useful and 40MHz channels appear to be currently default on most access points.

Combined channels need to be accounted for in channel planning. If you place two adjacent access points on channels 36 and 40 and enable 40MHz channels, the APs will end up taking turns. The AP on channel 36 will use channels 36–43 and the other will use 40–47. Due to the overlap they cannot transmit or receive at the same time. You should place them on channels 36 and 44 to account for this. In the standard the 40MHz channels are numbered 38, 46, 54… to avoid overlapping, but most user interfaces seem to use 20MHz numbering.

802.11ac introduced 80MHz and 160MHz channels. They have their own channel numbers as well, because extra wide channels are so easy to set to overlap. Using such wide channels makes even the 5GHz band crowded. Another problem is wait time for the channel availability. If there are other APs in the neighborhood our AP cannot transmit before all the channels are quiet at the same time. 802.11ac provides for dynamic channel width, which turns the setting to a maximum and the AP will choose the used channel width according to the environment. The extra wide channels also require support in the user device as well to be used. 160MHz channels are a Wave 2 feature so there is not much support at this point. In reality you cannot make a channel plan with just two channels so 160MHz should be reserved for point-to-point links where they really are useful.

The maximum transmission power set by the authorities is for the whole transmission. The maximums are calculated for 20MHz channels. The maximum should be halved (-3dBm) for 40MHz, quartered (-6dBm) for 80MHz and only one eighth (-9dBm) for 160MHz. Usually this doesn’t matter for access point use, since the maximums shouldn’t be used anyways. In point-to-point links it does matter and occasionally you need to concentrate the power on fewer channels to get a stable link. A double channel with the same nominal transmit power will also consume twice the electric power, which is important factor for mobile devices. If the wide channel will respectively increase the transmission speed (i.e. shorten transmission time) then it will cancel the increase in power consumption. In practice the power consumption will increase somewhat due to retransmissions.

20MHz channels are still very useful and often recommended. Especially in noisy environments or with lots of APs. Wide channels pick up noise on the whole channel width. Narrow channels pick up less noise. If there are many APs within reach, then it is better to assign each AP a separate channel. There are twice as many 20MHz channels than 40MHz channels.

Recommendations

Compatible user devices should be steered to 5GHz since there is more capacity and less interference. The simplest way is to turn down the transmission power of the 2.4GHz to the minimum or turn it off altogether.

40MHz channels on 5GHz are well supported and increase bandwidth. You just need to plan the channel use not to cause overlaps. Even if the original channel plan is perfect, DFS may cause unexpected channel switches causing overlaps. If you have a quiet environment and your devices support them, you may use 80MHz channels. One example of consideration is the fact that the lowest 80MHz will cover all non-DFS channels. If you want to use the wider channels you have to live with DFS restrictions. Don’t use 160MHz channels except for point-to-point or other special cases.

Related articles:

19 thoughts on “WiFi 5GHz band and wide channels”

  1. Thank you so much for all your generous sharing. It is a confusing channel landscape when contentious neighboring offices’ APs are also transmitting 40 Mhz widths, likely on the common “auto” channel selections – 40Mhz gets described on some analyzers and AP UIs as, say 36+40 – this means that the secondary bandwidth is 40 through 43, right? if not clear, then the AP will utilize 36 through 39, achieving 20 Mhz width?

    in troubleshooting and environment testing in a difficult RF setting, is it legitimate to set APs using 802.11ac to 20 Mhz widths, say, on channel 36 (occupying up to 39) and a nearby AP using the same BSSID at channel 40, occupying up to 43?
    just to assess coverage and connection stability?

    would certain clients that need to step back to 802.11n be happy with the 20Mhz channel width, and function, even if throughput performance was less than it could be?

    thanks again for all you do here.

    1. In 802.11n the administrator would set the channels width. If set to 40MHz then devices would always wait for both channels to clear before transmitting. In 802.11ac the administrator sets the maximum and the devices choose the channel width on a per-frame basis.

      The extension channel can be above or below the primary channel, there is no rule. However, most APs won’t let you choose. MikroTiks let you choose and with 80MHz channels it gets bewildering choosing between Ceee, eCee, eeCe and eeeC (as MT labels them). In theory it could make a difference since the clients can’t send on an extension channel only. The primary channel is mandatory and the extension channel has to be next to it. (I’m not getting into 80+80MHz channels, where the bonded 80MHz channels don’t have to be adjacent.)

      Yes, you can always use 20MHz channel width and often you should. With narrower channels you have more channels to choose from, which is great for dense installations, say a big lecture room with multiple APs. With narrow channels you also get less noise, better sensitivity and more transmit power (if you need it). There are many benefits. I’d say great many installations would do better with 20MHz channels than the default 40MHz. You should only use 40MHz for heavy file transfers or other high-bandwidth applications. Most aren’t.

      The disadvantages of wide channels become even more apparent with 80MHz channels or let alone 160MHz. This is very difficult to explain to customers, who believe they have paid for 5Gbps throughput.

  2. Thank you sir, my utmost gratitude for the effort of your
    team, i am a regular visitor to your website indeed, i thought to,
    (out of courtsey) provide some suggestion and feedback of my
    own , would be thankful if you could reply or acknowledge my recommendations to make this website more content focused
    .

    Regards.
    madden mobile hack

  3. Hi Petri,
    Really great posts on your site.
    I noticed one thing, which might not be correct, or I do not know well. You write that “The AP on channel 36 will use channels 36–43 and the other will use 40–47. ”
    I believe 40 MHz channel cannot span over channels 40+44 (40->47) only on 36+40 (36->43) OR on 44+48 (44->51).

    1. According to the standards you can choose your primary channel freely and specify the extension channel above or below the primary. You can easily create a mess. Many 802.11n systems let you choose any primary channel. Some systems let you specify the extension channel as well, but some set it for you (and didn’t tell). It gets even messier with wider channels where you could specify Ceee, eCee, eeCe or eeeC (e=extension, C=primary). Fortunately only MikroTik lets the user choose. That’s why the 802.11ac spec had the new channel numbers for wider channels. Very few vendors use them, though.

  4. So if we have less range on 5Ghz, Shall I keep one AP Cisco 3802i every 10-11 meters in my setup ? We have 65×60 meter floor and I ended up having 24 APs in that floor plan where we have corner closed office and in the middle, open cubicles typical office business floorplan.

    1. It is impossible to say based on the given information. It depends on many factors. How many users? How many devices? What kind of use? What kind of expectations?

      In general 10-11 meters is very tight. You’ll get an excellent network if you turn down the transmit power to the minimum, but the price is high. 15-20 meters apart is a more common choice for offices, depending on the construction. There are many kinds of cubicle walls, you know. If the workstations are wired and the Wi-Fi is used only for occasional smartphone data (not for VoIP calls) you could cover the floor with just four or five APs. It depends on the requirements. You can test this by placing a single AP at some location and walking around it and checking the signal strength. This is known as AP-on-a-stick because you often need some kind of support to get the AP to the ceiling.

      Don’t place the APs on a grid, think of a honeycomb instead – stagger the placements. Of course you need to make adjustments so that you get the APs close to the users if there are concentration areas. Still, do lower the transmit power. See 8 reasons to turn down the transmit power of your Wi-Fi

  5. Interesting perspective, about 5 Ghz having 6 db of implied attenuation (compared to 2.4 Ghz); But I have always considered it a dead heat because of the 6 db plus gain (per sq. ft.) of a 5Ghz antenna, over a 2.4 Ghz antenna. Of course the antenna gain is less
    When comparing your typical bad omni wips on a router.

    1. The 6dB difference comes directly from the Free Space Path Loss formula. Doubled frequency squared equals four times or quadruple loss. -3dB is half and -6dB is a quarter of the signal strength. I don’t think there is any room for argument in that.
      As for the antenna gains: The manufacturers try to keep the antenna gain as low as possible for omni antennas. A 6dB omnidirectional antenna for 5GHz would have less optimal antenna pattern. The manufacturer would also have to lower the transmit power accordingly to keep EIRP within limits. There are no advantages in higher gain omni antennas for multipurpose use.

    1. That depends on where you are. FCC has its own ruleset for the U.S., ETSI for Europe and NTT for Japan, to begin with. Note that there may be different limits for indoor and outdoor use. However, if you read my post on transmit power you will realize 25mW is plenty. (Unless you are building long distance point-to-point links, which is a different matter altogether.)

  6. For days I couldn’t figure out why my Sony android tv cant use 5Ghz band on mikrotik AP, your article pointed me to the right direction. Thanks.

  7. Hi, Great article. I have a question – I am using 5GHz wifi (ac; channel 36; 80MHz; on average 300Mbps); there are a lot of networks around and it seems to be the best option. However. I have an US router (im based in EU), which has UNII-3 channels. I am able to connect to e.g. 161 (it states its 80Mhz and not 20MHZ as you were stating above); the signal quality is excellent (way better than channel 36), but I get max 10Mbps (instead of 300Mbps on channel 36). I am not fully sure why?

    1. Most APs use the lowest channel as their primary channel. Then they may have 1, 3 or 7 secondary channels above the primary. However, this is not mandated and manufacturers have made different interpretations. If you have 80MHz on 161, then the only explanation is that you have three secondaries below. Running a primary on someone else’s secondary is very bad™. This would explain the poor performance. All APs gracefully share the primary channel by always listening first, but blatantly transmit over any secondaries. Your primary on 161 would be a victim.

      At least some (or most?) European devices won’t scan UNII-3 at all, so they won’t ever connect to your network. Just be aware of that.

  8. dynamic channel width seems an interesting concept but i wonder what would happen if two APs see that the environment is clear and start transmitting at the same time.
    I’m struggling to find sources with detailed info about this, do you have any recommendation ?

    1. If two devices transmit at the same time, a collision occurs. The devices detect the collision and back off. After a random length delay they try again and hopefully only one at a time. I would recommend the Sybex’ book on CWNA for more details.

  9. Got a question. I can’t find the answer anywhere. We have APs with 3 radios one of them can run as a sensor/spectrum analyzer. I started a spectrum capture and noticed some activity on channel 38 and 40 duty cycle about 15% if I remember correctly (not at work at the moment). This AP (our AP) is using channel 36 and looking at the RF information as seeing by this AP there is an external BSS broadcasting an SSID and it’s using 80 MHz and using channel 36. This office is located below an MRI machine but after researching I found that these machines don’t work in the same frequency.
    Is it possible that what I’m seeing in the spectrum be this AP using the 80 MHz bonded channels?
    Whatever this device it seems to be always active.
    I notice some spikes on channel 36 but it looks to be normal activity.
    what does an 80 MHz channel look like in the spectrum if using channel 36 as the primary channel?
    I have ekahau also but never seen something anything using channel 38.
    I can try to go in this office and see if ekahau produces the same results but in the meantime any info would be really appreciated.

    1. Bonded channels (40, 80 and 160MHz) have a primary channel, which is used for all control and management traffic (beacons, authentication etc.) The extra channels are only used for data and only if both devices so agree. Some devices may not support wide channels, so they only use the primary channel. Thus every frame will utilize the primary channel.

      If you are using the third radio to sensor the same frequency where the AP is transmitting on another radio, you will see a very strong signal, of course. The MRI machine could use 5GHz for some control or data transfer function unrelated to the actual imaging function.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.