Last week I wrote about the difference between a 2x2:2 and a 4x4:4 access point. The discussion that followed was more interesting than the article itself, as often!
The part that kept coming back was MIMO gain. Where does it come from, how much of it is really there, and what are you supposed to do with it when you sit down in Ekahau or Hamina and start setting transmit power?
That is the follow-up. Let us look at what those extra antennas actually do to the link, and then at the one place where ignoring them will quietly skew your design.
Where the MIMO Gain Actually Comes From
When we say a 4x4:4 AP has more MIMO gain than a 2x2:2 AP, we are compressing three different things into one term. They are worth separating, because they behave differently.
- Channel diversity and conditioning. Four transmit and receive chains mean four antenna elements, each with its own spatial and polarization response. The AP is not simply hearing the same signal four times. It is observing the channel from four different vantage points, which makes the channel matrix better conditioned and raises the MIMO gain that is available in the first place. Ben Toner made this point in the comments of the previous article, and he is right that this goes deeper than combining copies of a signal. Thank you Ben!
- Maximum Ratio Combining. On top of that diversity, the AP weighs and combines what each chain received before decoding. This is the receive-side gain most of us picture when we say MRC, and it improves the uplink SNR.
- Array gain on transmit. More chains also mean more total conducted power leaving the AP. This one is arithmetic rather than RF magic, and it is the one that shows up in your design tool.
One correction I want to carry over from that discussion. Transmit beamforming is not inherently a MIMO technology. It exists in MISO as well, with a single-antenna receiver. What it does is concentrate the transmission toward the receiver and improve the SNR at the client, which is valuable, but it is not the same mechanism as spatial multiplexing.
Four Chains, Two Streams: What Happens on the Air
This is the situation you will actually deploy: a 4x4:4 AP serving a fleet of 2x2:2 clients. The link is limited to two spatial streams, and no amount of hardware on the AP side changes that.
What changes is the quality of those two streams.
- Uplink. The client sends two spatial streams. The AP receives them across four elements with different spatial and polarization responses and combines them. The effective SNR goes up, the AP can decode a higher modulation, and each frame occupies less airtime.
- Downlink. The AP still sends two spatial streams, but it can map them across four chains and shape the transmission toward the client. The client sees a better SNR than it would from two chains.
- Neither of these adds a stream. The MCS index may climb, the retry count may drop, but the spatial stream count is fixed by the client.
And as I said in the previous article, the uplink gain and the downlink gain apply to different directions of the same conversation. They do not add together, and they never turn into the doubling that people expect.
The Place This Alters Your Design: Transmit Power
Here is where the theory stops being academic.
When you set a transmit power value on an AP in Ekahau or in Hamina, you have to know what that number refers to. Transmit power in Wi-Fi is expressed in three different ways:
- Per-chain transmit power. The power leaving each individual radio chain. Several vendors report power this way in their configuration interface.
- Total conducted power. The sum of the power leaving all the chains, before any antenna gain is taken into account. Doubling the number of chains adds 3 dB, so a four-chain AP puts out 6 dB more total conducted power than a single chain set at the same value.
- EIRP. The total conducted power plus the antenna gain. This is the number the regulator limits, and it is also the one that determines what the client actually receives.
Those three values describe the same radio at the same setting, and they are far apart. Set 15 dBm per chain on a four-chain AP and the total conducted power is 21 dBm, with the EIRP higher still once the antenna gain is added. Model that same AP as 15 dBm total instead and your predicted coverage will be several dB short of reality.
HPE Mist documents its own notation explicitly, and it is worth reading their page rather than assuming, because the convention is not universal.
The second half of the problem is the client. Your 4x4:4 AP can shout. A phone or a laptop with two chains and a small battery cannot. If you design your cell edge around the AP transmit power, you are designing around the strong half of the link and the uplink will fall short of your predictions. This is the AP and client power mismatch, and MIMO gain is precisely what makes it easy to miss: the AP hears well enough that the design looks fine until the client has to talk back at range.
How I Set Transmit Power for a Design
- Confirm the notation used by your design software before you type a number. Per chain or total, for that specific design tool. Check the documentation. As of today, both Ekahau and Hamina define the transmit power of the access point radio as total conducted power.
- Convert deliberately when you move between tools or vendors. The power calculator below does this well and removes the arithmetic mistake from the equation. This will help you understand how to configure your transmit powers in your configurations. https://power.potatofi.com/?from=hamina&to=openintent&apVendor=juniper&apModel=AP47&domain=us Thank you Joel!
- Design the cell edge on the client, not on the AP. Pick the least capable client that matters, use its realistic transmit power, and validate that the uplink closes at your target rate.
- Keep the link balanced rather than maximized. Turning AP power up because the hardware allows it widens the gap between downlink and uplink and buys you sticky clients instead of coverage.
- Then let the MIMO gain be a margin, not a plan. Better SNR, higher modulation, fewer retries, less airtime consumed. Take it as headroom on a design that already works without it.
One caveat. All of this assumes your clients are 2x2:2, which is what I see in the enterprise the vast majority of the time. If you are designing for a population of single-stream devices, old scanners, or IoT sensors, the same reasoning applies but the gap between AP and client widens further, and designing on the client side matters even more.
Resources
These are the references I keep going back to on this topic:
- Mist AP Transmit Power Notation. HPE Mist spells out exactly what its transmit power values mean and how they relate to the chains. Read this one first if you configure or model Mist APs, because it removes the guesswork the rest of the industry leaves you with.
- Access Points — Hamina Knowledgebase. The Hamina documentation for simulating access points, including the transmit power field.
- Don't forget about MIMO by Scott Lester. A short write-up on why the MIMO configuration of an AP has to be part of the transmit power conversation. It makes the per-chain versus total distinction concrete for anyone building a predictive design.
- Power Calculator by Joel Crane. A genuinely nice tool that converts transmit power values between vendors and between design tools. If you move a design from Hamina to another tool, or from one AP model to another, use this instead of doing the dB arithmetic by hand.
- HW041: Dealing with Client and AP Power Mismatch — Heavy Wireless, Packet Pushers. A full episode on the asymmetry between what an AP can transmit and what a client can transmit back. This is the deeper version of the second half of this article.