Resource Units from Wi-Fi 6 to Wi-Fi 8: What Actually Changed

Originally published October 2019 as a video walkthrough — rewritten and updated July 2026.

Back in 2019 I shot a quick video explaining Resource Units because Wi-Fi 6 had just dropped OFDMA on us and half the industry was nodding along pretending they understood it. Seven years, two generations of silicon, and a whole lot of stadium deployments later, RUs are still the single most important thing to understand about how modern Wi-Fi actually moves traffic. So let's do this properly — what RUs are, what changed in Wi-Fi 7, and what 802.11bn (Wi-Fi 8) is about to do with them.

What Are Resource Units, and Why Did OFDMA Need Them?

Before Wi-Fi 6, every transmission on a channel worked like one customer renting out an entire taco truck to order a single taco. Didn't matter if your frame was a 64-byte TCP ACK or a 4K video burst — you got the whole 20, 40, or 80 MHz channel to yourself for the duration of that transmission, and everybody else stood in line. In a house with six devices, fine. On the sand at Palm Tree Music Festival with 15,000 phones checking Instagram every eleven seconds? That line gets ugly fast.

OFDMA — Orthogonal Frequency Division Multiple Access — fixed that by letting the AP carve a single channel into smaller frequency chunks and serve multiple clients simultaneously in one transmit opportunity. Those chunks are Resource Units. Each RU is a block of subcarriers (the individual ~78.125 kHz-wide tones that make up an OFDM channel), and the AP's scheduler hands them out like a host seating parties at tables sized to the group. Two people don't get the twelve-top anymore.

The key mental shift: with OFDMA, airtime efficiency stops being about raw speed and starts being about scheduling. The AP becomes traffic control. That's why Wi-Fi 6 was never really a "faster" standard — it was a denser standard. Having built high-density networks from World Surf League events to the McAllen citywide deployment, I can tell you the difference between an AP that schedules RUs well and one that doesn't is the difference between a network that survives halftime and one that doesn't.

What Are the RU Sizes in Wi-Fi 6?

802.11ax defined a fixed menu of RU sizes, named by their tone (subcarrier) count. Here's the full table, including the 320 MHz-wide addition that arrived with Wi-Fi 7:

RU type	Subcarriers (tones)	~Bandwidth	Typical use
26-tone RU	26 (24 data + 2 pilot)	~2 MHz	Tiny payloads — IoT sensors, VoIP, TCP ACKs; up to 9 clients at once in 20 MHz
52-tone RU	52 (48 data + 4 pilot)	~4 MHz	Light traffic, voice, small transactional flows
106-tone RU	106 (102 data + 4 pilot)	~8 MHz	Medium payloads, mixed client scheduling
242-tone RU	242 (234 data + 8 pilot)	~20 MHz	One client gets a full 20 MHz channel's worth
484-tone RU	484 (468 data + 16 pilot)	~40 MHz	High-throughput clients in 40 MHz+ channels
996-tone RU	996 (980 data + 16 pilot)	~80 MHz	Video, big downloads, single fast client
2×996-tone RU	1,992	~160 MHz	Maximum Wi-Fi 6E single-client allocation
4×996-tone RU (Wi-Fi 7)	3,984	~320 MHz	Full-fat 802.11be channel in 6 GHz

The catch in Wi-Fi 6 — and this is the part that mattered in the field — was the one-RU-per-client rule. Each client got exactly one RU per transmission. If the scheduler handed a client a 106-tone RU and there was a stray 26-tone RU sitting next to it unassigned, that spectrum just sat there. Wasted airtime, wasted capacity. Cisco Meraki's Wi-Fi 7 technical guide calls this out directly: portions of spectrum "might be left unused and be wasted." In dense environments, those crumbs add up to whole meals.

How Did Multi-RU Change the Game in Wi-Fi 7?

Wi-Fi 7 (802.11be) killed the one-RU-per-client rule with Multi-RU (MRU): a single client can now be assigned multiple RUs in the same transmission. The standard defines two families of combinations:

• Small-size MRUs — contiguous combos like 26+52 tones (~6 MHz) and 26+106 tones (~10 MHz), which let the scheduler sweep up those leftover crumbs within a 20 MHz segment instead of stranding them.
• Large-size MRUs — combos like 484+996 (~120 MHz) and 996+484 variants up through 3×996+484, which can be non-contiguous. This is where it gets interesting.

Non-contiguous large MRUs are what make preamble puncturing genuinely useful. Puncturing lets an AP notch out a slice of a wide channel that's being trashed by interference — say, a 20 MHz chunk of your 160 MHz channel that's sitting on top of a radar hit or a neighbor's legacy network — and keep transmitting on everything around the hole. Wi-Fi 6 had a limited version of this; Wi-Fi 7 made punctured transmission a mandatory, standardized capability, and MRU is the mechanism that lets a client actually use the surviving spectrum on both sides of the notch as one logical allocation. RUCKUS has a solid write-up on how punctured transmission works in practice.

Field translation: before Wi-Fi 7, one bad 20 MHz neighbor could force you to abandon a wide channel entirely — drop from 160 to 80, eat the capacity loss, move on. Now you punch a hole in the channel and keep 140 MHz of it working. For anyone deploying wide channels in 5 GHz DFS territory or crowded 6 GHz, that's not a spec-sheet feature, that's a design tool.

What Does Wi-Fi 8 Do With Sub-Channels?

802.11bn — Wi-Fi 8, branded Ultra High Reliability (UHR) — is the first generation in a long time that isn't chasing a bigger throughput number. Per HPE's Wi-Fi 8 overview, it keeps Wi-Fi 7's ceiling intact: 320 MHz channels, 4096-QAM, 8 spatial streams, MLO, and the same RU/MRU structure underneath. What it changes is how flexibly that spectrum gets used when conditions are messy — which, in the real world, is always.

The headline for this article is Dynamic Sub-channel Operation (DSO). Today, if your AP runs a 160 MHz channel, a client that only supports 80 MHz — or one sitting far enough away that wide channels aren't viable — parks on the primary channel along with every other constrained client, and the primary becomes a congestion chokepoint while secondary spectrum idles. DSO lets the AP dynamically steer individual clients onto different sub-channels of the wide channel on a per-transmission basis, without the client re-associating or the AP re-plumbing its channel plan. Its sibling feature, Non-Primary Channel Access (NPCA), lets devices contend and transmit on secondary channels when the primary is busy — something that was heresy in every previous generation. The MAC-level mechanics are being hammered out in the IEEE 802.11bn task group documents right now, with ratification expected around 2028 and first chipsets already previewed in 2026.

Think of the progression this way: Wi-Fi 6 taught the AP to seat multiple parties in the dining room at once. Wi-Fi 7 let one party push tables together, even around a support column. Wi-Fi 8 lets the host move you to a different dining room mid-meal because a tour bus just pulled up outside yours. Same restaurant, radically more flexible seating.

What Does This Mean for Your Designs?

Twenty-plus years of doing this tells me the spec only matters where it changes what you build. Here's where RU evolution actually touches your designs:

• Wide channels are back on the menu. With mandatory puncturing and MRU in Wi-Fi 7, an 80 or 160 MHz channel with one dirty slice is no longer disqualified. Survey for interference, but design around it instead of retreating from it.
• High density is a scheduler game. RU efficiency lives in AP firmware. Two APs with identical spec sheets can behave completely differently under load — test vendors with real concurrent clients, not iPerf from one laptop.
• Small RUs are your IoT strategy. Hundreds of sensors sipping 26-tone RUs coexist beautifully with big flows on large MRUs — if your airtime fairness and QoS policies are set up to let the scheduler do its job.
• Don't wait on Wi-Fi 8 for capacity. 11bn adds reliability, not speed. If you need capacity today, Wi-Fi 7 with clean channel planning gets you there. If you're spec'ing for 2028+, leave room for multi-AP coordination in your controller architecture.
• Mixed-client reality still rules. Every one of these features requires client-side support. Your festival crowd in 2026 is still full of Wi-Fi 5 and Wi-Fi 6 devices, so design your primary-channel capacity for the worst client, not the best.

What Are the Questions I Keep Getting About RUs?

Does OFDMA actually help on a small network? Honestly, not much. With a handful of clients, the taco truck line was never long. RUs earn their keep in density — offices, venues, stadiums, anywhere dozens of clients hit one radio. On the home network, MLO and wide channels matter more than RU scheduling ever will.

Do I need to configure RU sizes myself? No — and you can't. RU allocation is decided per-transmission by the AP's scheduler. What you control is everything that feeds it: channel width, QoS policy, band steering, client mix. Get those right and the scheduler eats well.

Will Wi-Fi 8 make my Wi-Fi 7 deployment obsolete? No. Wi-Fi 8 keeps the same PHY ceiling — same channels, same modulation, same RU structure — and adds reliability mechanisms on top. A well-designed Wi-Fi 7 network deployed today will still be a well-designed network when 11bn gear ships around 2028. RF fundamentals don't expire with the standard revision.

Frequently asked

Does OFDMA actually help on a small network?

Not much. With a handful of clients there's little contention for RU scheduling to solve. RUs earn their keep in high-density environments — offices, venues, stadiums — where dozens of clients share one radio. On home networks, MLO and wide channels matter more than RU scheduling.

Do I need to configure RU sizes myself?

No — RU allocation is decided per-transmission by the AP's scheduler and isn't user-configurable. What you control is everything that feeds it: channel width, QoS policy, band steering, and client mix. Get those right and the scheduler does its job well.

Will Wi-Fi 8 make my Wi-Fi 7 deployment obsolete?

No. Wi-Fi 8 (802.11bn) keeps the same PHY ceiling as Wi-Fi 7 — 320 MHz channels, 4096-QAM, the same RU structure — and adds reliability features like Dynamic Sub-channel Operation and multi-AP coordination on top. A well-designed Wi-Fi 7 network deployed today remains solid when 11bn gear ships around 2028.

The original 2019 video walkthrough is still up: youtu.be/CYBlbYrsFHI