Wi-Fi 8 - IEEE 802.11bn Ultra-High Reliability

802.11bn retains the same physical foundation as Wi-Fi 7: 2.4 GHz / 5 GHz / 6 GHz bands (1–7.25 GHz), maximum 320 MHz channel bandwidth, 4096-QAM modulation, up to 8 spatial streams per link, and ~23 Gbps theoretical peak. The peak rate does not increase. The mission is making every percentile of performance better - not raising the ceiling, but raising the floor.

Problem: In Wi-Fi 6/7, OFDMA Resource Units are contiguous subcarrier blocks. In the 6 GHz band under LPI rules, the AP can transmit at up to 30 dBm EIRP - but a phone using a small contiguous 26-tone RU is PSD-limited to much lower power per tone, capping its uplink range.

Solution (DRU): Scatter the same number of subcarriers across the entire available distribution bandwidth instead of in a contiguous block. Each subcarrier can now transmit at the PSD limit independently. Net effect: the same information content transmitted with dramatically higher total power - doubling or tripling the reliable uplink range for the same device in the 6 GHz band.

Problem (link budget imbalance): An AP can transmit at up to 23 dBm (indoors) or 36 dBm (outdoors with AFC). A phone transmits at ~20 dBm. That 10+ dB asymmetry means the AP can hear the phone from 50 m away, but the phone cannot reliably respond - uplink fails before downlink. This is the fundamental range problem for all Wi-Fi deployments.

Solution (ELR): A dedicated 20 MHz-only PPDU format with BPSK and QPSK at two coding rates - 1.67 Mbps and 3.33 Mbps - that sacrifices throughput for extreme link robustness. ELR uses 2xLDPC and extended preamble for additional coding gain. The ELR-MARK field in the preamble carries BSS Color for early OBSS filtering.

Prior Wi-Fi: all spatial streams in a MU-MIMO or beamforming transmission use the same MCS (e.g., all 4 streams at MCS 11 1024-QAM). UEQM (Unequal Modulation) allows each spatial stream to use a different modulation order based on that stream's specific channel quality. A beam pointing toward a client through a concrete wall uses QPSK on its weakest stream while using 1024-QAM on its strongest. Net effect: significantly better beamforming performance in complex multi-antenna environments with heterogeneous link qualities.

802.11bn introduces additional MCS levels to close the SNR gaps between existing MCS indices. In Wi-Fi 6/7, there are step jumps at certain SNR values where the link could theoretically support a higher rate but no MCS index exists for that operating point. New MCS fill these gaps, enabling finer-grained link adaptation and better throughput at intermediate SNR values. Result: smoother rate-vs-range curves and fewer retransmissions at SNR boundary conditions.

Low-Density Parity Check (LDPC) has been in 802.11 since 802.11n as an optional coding scheme. 802.11bn upgrades LDPC to a two-pass (2x) implementation that provides additional coding gain - particularly useful for ELR frames and for critical-reliability data streams where even a 1 dB improvement in effective SNR matters. 2xLDPC is the error correction equivalent of what ELR is to the PHY format: sacrificing some overhead for measurable reliability gains at the receiver.

Problem: In all prior 802.11 standards, the primary 20 MHz channel carries all MAC signalling - beacons, probe requests, trigger frames, channel announcements. When this channel is busy (often due to OBSS activity), the entire 80/160/320 MHz channel is blocked even if the other 14 non-primary 20 MHz sub-channels are clear.

Solution: NPCA allows an AP or STA to temporarily designate a different 20 MHz channel as the primary channel (the "NPCA primary channel") when the original primary is occupied. The AP broadcasts NPCA parameters including the candidate NPCA primary channel, switching delays, and the conditions under which NPCA is permitted. The STA also declares its own switching delay. The result: congested primary channels no longer block the full wideband operation.

Problem: A 320 MHz AP serving a 160 MHz client wastes 160 MHz. A 160 MHz AP serving an 80 MHz client wastes 80 MHz. There was previously no mechanism to dynamically shift a STA's operating bandwidth to better utilise the AP's wider channel. Under-utilised spectrum in one part of the band while another part is congested.

Solution: DSO allows the AP to dynamically allocate frequency resources to a STA operating outside the STA's normal bandwidth. For example: move an 80 MHz STA into a different 80 MHz slice of the AP's 320 MHz channel to reduce congestion in the primary slice. Works in conjunction with NPCA for comprehensive spectrum flexibility.

Related to DSO. DBE allows the AP to expand a STA's operating bandwidth when spectrum becomes available - the inverse of DSO. A STA negotiated at 80 MHz can be temporarily expanded to 160 MHz during periods of low channel load, then returned to 80 MHz when congestion increases. This maximises throughput during favourable conditions without sacrificing compatibility.

P-EDCA (Prioritised EDCA): A STA with an urgent frame sends a Defer Signal (DS-CTS) - a new control frame type - that temporarily excludes all STAs with non-urgent traffic from channel contention. This clears the airwaves for one high-priority TXOP. The DS-CTS is the Wi-Fi 8 equivalent of a 5G QoS priority class - it carves out a guaranteed transmission slot.

Preemption: If an urgent frame arrives during an ongoing transmission of lower priority, 802.11bn allows stopping that transmission and starting the urgent one. The interrupted transmission resumes after the high-priority frame is acknowledged. This is conceptually borrowed from 5G pre-emptive scheduling and is new to 802.11.

Combined latency target: P-EDCA + preemption target sub-10 ms worst-case latency for AR/VR, industrial control, and mission-critical signalling.

DPS reduces power consumption during link listening periods. An AP in DPS mode can reduce its antenna activity and receiver sensitivity during intervals when no associated STA is expected to transmit (based on TWT schedules and traffic patterns). Designed for APs in battery-powered or energy-constrained deployments. Complements TWT at the STA side - when the STA is asleep (TWT), the AP side can also save power rather than maintaining full receive readiness.

What it solves: Even with 802.11r Fast Transition, there is a handover latency - the STA disassociates from AP1 before completing reassociation on AP2. In a 50 ms window, a voice call drops a packet, an XR frame is missed, or an industrial control message is delayed. SMD eliminates this by redefining the roaming event as a link transition within a maintained association rather than a disassociation and new association.

How it works: STAs in a Seamless Mobility Domain maintain association with multiple APs simultaneously via MLO (from 802.11be). When a STA moves from AP1 to AP2, it uses a dedicated link on AP2 to complete key re-establishment before dropping the link to AP1. The data path moves from AP1 to AP2 with zero packet loss and zero latency interruption. Security context (PMK, PTK) is maintained throughout - there is no EAPOL 4-Way Handshake re-run. SMD roaming is measured in microseconds, not milliseconds.

A Bounded ESS is a defined set of APs that form a single coordinated wireless domain with strict membership rules. Within a Bounded ESS, MAPC, SMD, RSNO (secure roaming), and coordinated resource allocation all operate under a unified policy. The boundary is administratively defined - only APs within the Bounded ESS participate in coordination. STAs inside a Bounded ESS get full Wi-Fi 8 benefits; STAs outside get standard 802.11 treatment. This is the architectural unit of Wi-Fi 8 deployment - analogous to a 5G cell cluster or a coordinated RAN cluster.

Key insight from David Coleman (Extreme Networks, ExtremeConnect 2026): "Your Wi-Fi infrastructure is the most distributed compute platform you already own - edge AI turns it into a business application platform." The 802.11bn reliability improvements (sub-millisecond latency, 25% less packet loss, MAPC coordination) are what make this viable - AI inference results are only useful if the network can act on them in real time.

The three features most likely to define Wi-Fi 8 adoption in practice are: (1) MAPC / Co-TDMA - this is the feature that enables industrial and mission-critical wireless for the first time; (2) SMD seamless roaming - this is what XR and autonomous mobile devices need; (3) DRU uplink - this solves the IoT uplink range problem that TWT alone cannot address. WPA3 mandatory is not a headline feature but is the most operationally disruptive change for organisations still running WPA2-only infrastructure.