Over the last couple of years, I have watched 6 GHz cells come in short at validation on designs I was confident in.
The design looked right, the APs were installed where we planned them, and the validation survey still showed the 6 GHz signal falling a little short of what we expected. Clients that could use 6 GHz were also not landing on it as often as they should have.
So we changed our default. We now set the 6 GHz transmit power 3 dB higher than the 5 GHz transmit power on the same AP. If an AP runs at 11 dBm on 5 GHz, it runs at 14 dBm on 6 GHz. It depends on the environment, but it has consistently brought us closer to the outcome we design for: a good user experience and a healthy share of capable clients on 6 GHz.
Let us look at why the 6 GHz cell appears smaller in the first place, and why 3 dB is the right size of correction rather than more. Everything here is about Low Power Indoor (LPI) in Canada and the US, which is what I deploy today. Standard Power is a different conversation, and I will cover it in a separate post next week.
The Rules That Frame the 6 GHz Cell
Before talking about physics, it helps to know what LPI actually allows, because the rules are written differently from 5 GHz. In Canada, RSS-248 sets the limits, and they match the FCC rules in the US:
- LPI access points: 5 dBm/MHz maximum EIRP spectral density, and 30 dBm maximum EIRP.
- Clients connected to an LPI access point: −1 dBm/MHz maximum EIRP spectral density, and 24 dBm maximum EIRP. The client is always 6 dB below the AP.
Indoors, the limit that matters is the power spectral density (PSD), not the 30 dBm ceiling. Because the limit is per MHz, the maximum EIRP grows with the channel width:
- 20 MHz: 18 dBm for the AP, 12 dBm for the client
- 40 MHz: 21 dBm for the AP, 15 dBm for the client
- 80 MHz: 24 dBm for the AP, 18 dBm for the client
- 160 MHz: 27 dBm for the AP, 21 dBm for the client
Two consequences are easy to miss. First, a wider channel gives the AP more total power, but not more power per MHz, so the SNR on each subcarrier does not improve. Wider channels are a capacity tool, not a range tool. Second, beacons and probe responses are sent on the 20 MHz primary channel, so the frames a client uses to discover and judge a 6 GHz BSS are held to the 20 MHz limit, whatever the width of the channel.
Region matters here. In Europe, Commission Implementing Decision (EU) 2021/1067 limits LPI devices to 23 dBm mean EIRP and 10 dBm/MHz, and the same limits apply to LPI access points and LPI clients. There is no 6 dB penalty on the client. Because 10 dBm/MHz over 20 MHz already reaches 23 dBm, the ceiling is hit on a 20 MHz channel, and a wider channel brings no additional EIRP: it spreads the same 23 dBm thinner. On a 20 MHz channel, a European AP is allowed 5 dB more than a Canadian one, and a European client 11 dB more.
A European engineer starts from a very different link budget, so a North American 6 GHz power plan should never be copied to a European site, or the reverse.
Why the Same Power Lands Weaker at 6 GHz
With the AP at the same transmit power on both bands, the 6 GHz signal arrives at the client weaker than the 5 GHz signal. Several things contribute, and they add up:
- Free space path loss. The radio wave does not fade faster at 6 GHz. What changes is the receiving antenna: at the same gain, its effective capture area shrinks with the square of the wavelength, so it collects less energy at higher frequencies. You can run these numbers for your own channels and distances with our FSPL calculator.
- Building materials. Attenuation through walls generally increases with frequency, and the effect is strongest in dense materials such as concrete and brick. Light partitions change little, which is why open spaces hide the problem and walled environments expose it.
- Antennas at both ends. Tri-band AP antennas and the small antennas inside phones and laptops are not equally efficient across 2.4, 5, and 6 GHz. Check the AP datasheet for the 6 GHz antenna gain rather than assuming it matches 5 GHz. Client antenna performance is rarely published, which is one more reason to validate with real devices.
- Channel width. We tend to deploy wider channels on 6 GHz because we have so many of them. At the same total transmit power, every doubling of the channel width spreads the energy over twice the spectrum and costs 3 dB of SNR per subcarrier. This is not a propagation effect, but in the field it looks exactly like one.
None of these is dramatic on its own. Together, they explain why a 6 GHz cell built at 5 GHz power levels comes in short at validation, even when the design was right.
Why 3 dB, and Not the Maximum
If 6 GHz lands weaker, why not push it toward its LPI limit? Because the client still has to answer, and on LPI the client is regulated harder than the AP.
Let us work the example through. The 11 dBm and 14 dBm values are transmit power, the number you type into the controller or the design tool, before the antenna. The rules are written in EIRP, so the antenna gain has to be added first:
- EIRP = transmit power + antenna gain
- EIRP spectral density = EIRP − 10 × log10 of the channel width in MHz, which means subtracting 13 dB for 20 MHz, 16 dB for 40 MHz, and 19 dB for 80 MHz
Depending on how your vendor reports power, you may also need to account for the MIMO gain across the transmit chains, as I covered in my article on MIMO gain. To keep the arithmetic simple, let us assume a 6 GHz antenna gain of 5 dBi and leave the MIMO gain aside. Check the datasheet of your own AP, because this value changes from one model to the next. With 14 dBm of transmit power, the AP radiates 19 dBm EIRP:
- On a 20 MHz channel, 19 dBm EIRP is about 6 dBm/MHz. That is above the LPI limit of 5 dBm/MHz, so the AP is held to 18 dBm EIRP, which is 13 dBm of transmit power with this antenna. It is also 6 dB above the most a client is allowed to transmit on that channel (12 dBm).
- On a 40 MHz channel, 19 dBm EIRP is about 3 dBm/MHz, under the AP limit, and 4 dB above the client limit of 15 dBm.
- On an 80 MHz channel, 19 dBm EIRP is about 0 dBm/MHz, and 1 dB above the client limit of 18 dBm.
Two things stand out. On 20 MHz channels, our default already sits at the LPI ceiling, so there is no headroom left to add. And on every channel width, the downlink is already louder than anything an LPI client is allowed to send back.
Pushing the 6 GHz power higher on 40 or 80 MHz channels, where the rules would still allow it, would only widen that gap and recreate the asymmetric link I described in my previous article on transmit power, except this time the regulator guarantees it. That is why 3 dB is the right size of correction, and not more.
The other half of the outcome is client distribution. A 6 GHz-capable client discovers the 6 GHz BSS through the Reduced Neighbor Report in the 2.4 GHz and 5 GHz beacons, or by scanning the preferred scanning channels, then decides where to connect largely on signal strength. When the 6 GHz beacon arrives a few dB weaker than the 5 GHz one, many clients simply stay on 5 GHz. Some client vendors, Zebra for example, document a 6 GHz preference that tolerates a slightly lower 6 GHz signal, but you cannot count on every client doing the same. Bringing 6 GHz closer to 5 GHz at the client is what moves those clients over.
How I Now Design 6 GHz
- Treat 6 GHz as its own band. Set the 6 GHz power independently in Ekahau or Hamina, rather than reusing the 5 GHz plan with more channels.
- Choose the channel width before the power. The width decides your power per MHz and the limits that apply to both the AP and the client.
- Start at the 5 GHz power plus 3 dB. Then adjust for the environment: less in open, dense spaces where cells already overlap, more in walled spaces where the materials take their share.
- Convert transmit power to EIRP per MHz before comparing with the rules. Add the antenna gain, subtract 10 × log10 of the channel width, then compare with 5 dBm/MHz for the AP and −1 dBm/MHz for the client.
- Validate on 6 GHz specifically. Survey with a 6 GHz-capable adapter, and measure wall attenuation at 6 GHz rather than reusing 5 GHz values.
- Look at the uplink, not only the coverage map. Check the client RSSI and MCS seen at the AP, and the uplink retries, especially on 20 MHz channels.
- Watch where the clients land. After deployment, compare the number of 6 GHz-capable clients with the number actually connected on 6 GHz. That ratio tells you whether the power plan is doing its job.
The 3 dB is a default, not a constant. In a dense space with short distances to the clients, the right offset may be smaller. In a building full of concrete, it may need to be larger. And all of this changes with Standard Power, where the AP and client limits are much higher, which is the subject of next week's post.
Resources
If you want to go deeper on the rules and the trade-offs behind them, these are worth your time:
- RSS-248 — Radio Local Area Network (RLAN) Devices Operating in the 5925-7125 MHz Band, ISED. The Canadian rules themselves. Sections 4.5.2 and 4.5.3 hold the LPI access point and client limits, and reading them once is better than relying on secondhand tables, including the ones above.
- 6-GHz Unlicensed Spectrum Regulations and Deployment Options White Paper, Cisco. A clear overview of the power classes, the LPI requirements such as integrated antennas and indoor-only operation, and how AFC fits in. Start here if the regulatory side is new to you.
- Why we need 6GHz Standard Power and AFC, Cisco Community. Explains why beacons stay limited to 18 dBm on LPI and why high ceilings push LPI to its edge. A good bridge to the Standard Power discussion.
- Settle the Channel Width Before the Transmit Power, WLAN Professionals. Walks through LPI and Standard Power client power channel width by channel width. Read it after the Cisco white paper.
- Electromagnetic Signal Attenuation in Construction Materials, NIST, 1997. A laboratory study that measured how brick, masonry block, several concrete mixes, reinforced concrete, glass, plywood, lumber, and drywall attenuate signals from 0.5 to 8 GHz, at several thicknesses. It is old, but it covers 2.4, 5, and 6 GHz in a single dataset, which makes the frequency trend behind the building materials point easy to see.
Have you also had to run 6 GHz above 5 GHz in your LPI deployments, and if so, by how much?