| Why sounding | Beamformer needs channel state (H matrix per subcarrier) to compute steering matrix Q. 802.11n-era implicit sounding (reciprocity) abandoned → everything since 11ac uses EXPLICIT feedback. | No fresh sounding = stale steering = beamforming loss instead of gain. Sounding cadence ~10-100 ms per client, costs airtime. |
| VHT/HE/EHT sounding sequence | NDPA (lists STA AIDs + feedback type) → SIFS → NDP (null data packet: preamble/LTFs only, no data) → SIFS → BF Report from first STA → BF Report Poll → next STA report … | HE/EHT: reports can be pulled via Trigger (BFRP) in UL OFDMA - parallel feedback. Wireshark: type_subtype 37 (NDPA), Action VHT/HE categories for reports. |
| Compressed BF report | Feedback = per-subcarrier-group Givens-rotation angles (φ, ψ) quantized (SU: 4+2 or 6+4 bits; MU: 7+5 / 9+7) + per-group SNR. Ng grouping (2/4/16) trades size vs resolution. | MU feedback is BIG (tens of kB at 160 MHz 4SS) - why MU-MIMO gains vanish for fast-moving clients. See Advanced PHY Mathematics sheet for the Givens math. |
| CQI feedback (HE) | Per-RU channel quality report variant for OFDMA scheduling decisions. | Scheduler input, not steering. |
| DL MU-MIMO flow | Sounding → AP computes zero-forcing/MMSE precoding → MU PPDU with per-user spatial streams (HE-SIG-B/EHT-SIG carries user list) → each STA replies BA (staggered or triggered). | User pairing requires spatially separable clients (low correlation) + similar MCS-capable SNRs. Max 4 users/group (11ac wave2 style), 8 (11ax DL). |
| UL OFDMA flow (11ax/be) | BSRP Trigger → STAs report queue depth (BSR) → AP builds schedule → Basic Trigger (per-STA: RU, MCS, target RSSI, length) → STAs TX TB-PPDUs SIMULTANEOUSLY, pre-corrected in time/power/frequency → AP sends Multi-STA BlockAck. | Uplink becomes scheduled, cellular-like. STAs must hit AP within ±0.4 µs timing and tight power window - why UL OFDMA misbehaves with weak/distant clients. Trigger fields readable in Wireshark (wlan.trigger.*). |
| MU-RTS/CTS | MU-RTS Trigger to group → simultaneous CTS from all → NAV protection for the MU exchange. | Protects long MU TXOPs; also EMLSR link-switch primer in 11be. |
| MU EDCA | After being triggered, STA switches to MU EDCA parameters (usually much less aggressive) for MU EDCA Timer duration → discourages contention-based UL while AP is scheduling. | Badly tuned MU-EDCA = "11ax uplink slower than 11ac" complaints. Check Ext IE 38 values. |
| UORA | UL OFDMA Random Access: Trigger allocates RA-RUs (AID 0/2045); unassociated or unpolled STAs contend via OFDMA backoff (OBO) inside them. | Used for assoc requests/BSR from unscheduled STAs; niche client support. |