Dynamic Subband Operation in Wi‑Fi 8

• Dynamic Subband Operation (DSO) is a MAC-layer mechanism that recurrently slices the spectrum and reassigns subbands to different STAs based on current channel, queue, and delay conditions.
• The method uses a weighted bipartite matching approach, integrating CSI, queue backlog, and delay penalties to optimize throughput and manage interference per control interval.
• Simulations in Kom8ndor demonstrate that DSO achieves up to 32% higher throughput over legacy schemes with 99.99% reliability and 5 ms 99th-percentile latency in dense deployments.

Searching arXiv for the specified paper and closely related 802.11bn/Komondor work to ground the article in current literature. Dynamic Subband Operation (DSO) in IEEE 802.11bn denotes the ability of an AP, or a coordinating multi-AP cluster, to split its operating spectrum into multiple smaller subbands and dynamically assign each subband to a different STA or group of STAs on a per-beacon or per-TWT-slot basis. In the Kom8ndor simulator, which extends the open-source Komondor platform with 802.11bn features, DSO is presented alongside MAPC and NPCA as part of the Wi-Fi 8 feature set oriented toward Ultra-High Reliability (UHR) (Wilhelmi et al., 24 Jun 2026). The mechanism is intended to maximize aggregate throughput by exploiting frequency diversity, enforce UHR and bounded latency by isolating worst-case interferers into different subbands, and preserve coexistence with legacy devices by reverting transparently to full-band operation when necessary (Wilhelmi et al., 24 Jun 2026).

1. Position within IEEE 802.11bn and Kom8ndor

The forthcoming IEEE 802.11bn amendment is described as a paradigm shift in Wi-Fi, with ambitious performance targets under UHR. Within that setting, Kom8ndor is introduced as a discrete-event network simulator for Wi-Fi 8 and beyond, extending Komondor with 802.11bn-oriented functionality. Among the newly added features, the work highlights Multi-Access Point Coordination, Non-Primary Channel Access, and Dynamic Subband Operation (Wilhelmi et al., 24 Jun 2026).

Within this simulator context, DSO is not a static channelization mechanism. The defining distinction from traditional Wi-Fi is that conventional APs statically partition 20/40/80/160 MHz channels into sub-channels only at association or device provisioning time, whereas under DSO the partitioning and subband-to-STA mapping can change every control interval, such as every 100 ms beacon or per TWT service period (Wilhelmi et al., 24 Jun 2026). This makes DSO a MAC-driven spectrum assignment procedure rather than a one-time spectrum layout.

A common misconception is to treat DSO as merely narrower-channel operation. The formulation used in Kom8ndor instead frames DSO as recurrent spectrum slicing and reassignment. This suggests that its novelty lies less in subbandization itself than in the dynamic control loop that binds CSI, queue state, and delay-awareness to subband assignment (Wilhelmi et al., 24 Jun 2026).

2. Formal model and optimization objective

The paper defines the total available bandwidth as $B_{tot}$, for example $320$ MHz in the $6$ GHz band, partitioned into $M$ subbands indexed by $m = 1 \ldots M$, with subband widths $W_m$ such that $\sum_m W_m = B_{tot}$ (Wilhelmi et al., 24 Jun 2026). The associated STA set is denoted $U$, with cardinality $|U| = U$. For each STA-subband pair $(u,m)$ at time $320$0, the model uses the complex channel gain $320$1, subband transmit power $320$2, and noise PSD $320$3 (Wilhelmi et al., 24 Jun 2026).

The achievable PHY rate is given as

$320$4

Every control interval, DSO solves for an assignment matrix $320$5 with binary variables $320$6 and objective

$320$7

subject to

$320$8

$320$9

$6$0

These constraints enforce that each subband is assigned to at most one STA, optionally cap the number of subbands per STA, and optionally impose reliability or delay constraints (Wilhelmi et al., 24 Jun 2026). The implementation further augments the objective with per-STA weights $6$1, for example based on queue backlog or deadline-aware priority, yielding

$6$2

In this form, DSO can be read as a weighted scheduling problem over subband-frequency resources. A plausible implication is that the method unifies link adaptation and queue-aware scheduling at the granularity of subbands rather than full channels, which is consistent with the stated UHR and bounded-latency objectives (Wilhelmi et al., 24 Jun 2026).

3. Runtime procedure in Kom8ndor

The implementation follows a recurring DSO period $6$3, which may be equal to the beacon interval or to a configured TWT service-period boundary (Wilhelmi et al., 24 Jun 2026). At AP setup, the total band $6$4 is split into $6$5 equal-width subbands of size $6$6, and a DSO context is created containing the subband list, initial CSI placeholders, and an initial assignment that can flatten to full-band if no CSI is available (Wilhelmi et al., 24 Jun 2026).

The runtime reconfiguration loop is specified in four stages. First, each STA periodically reports per-subband CQI or measured SNR through uplink feedback elements carried in data or control frames. Second, the AP computes a scheduling metric for each pair $6$7:

$6$8

Here $6$9 is the queue backlog, $M$0 encodes closeness to deadline, and $M$1 are tunable weights (Wilhelmi et al., 24 Jun 2026). Third, assignment optimization is performed as a weighted bipartite matching, either exactly through Hungarian or heuristically through greedy sorting by $M$2 (Wilhelmi et al., 24 Jun 2026). Fourth, the internal MAC state is updated so that subsequent data and ACK exchanges on subband $M$3 are destined to STA $M$4 if $M$5 (Wilhelmi et al., 24 Jun 2026).

Control signaling is carried in the next beacon frame via a DSO Information Element encoding the number of subbands $M$6 and their center frequencies, per-subband STA association IDs, and the timing offset indicating when the new mapping takes effect (Wilhelmi et al., 24 Jun 2026). Legacy STAs ignore the DSO IE and continue on the full band, while multi-AP clusters can coordinate beacon schedules to avoid inter-AP collisions (Wilhelmi et al., 24 Jun 2026).

This architecture makes DSO operationally dependent on periodic CSI acquisition, optimization, and beacon-based dissemination. It also clarifies that backward compatibility is achieved through selective interpretation of the DSO IE rather than through separate legacy signaling paths.

4. Simulation configuration and operating regime

The evaluation uses a topology of four APs arranged in a $M$7 m$M$8 plaza, each with $M$9 MIMO radios, and sixteen uniformly scattered STAs, each generating uplink and downlink traffic (Wilhelmi et al., 24 Jun 2026). The radio setup assumes total bandwidth $m = 1 \ldots M$0 MHz in the $m = 1 \ldots M$1 GHz band, with subband granularity $m = 1 \ldots M$2 and therefore $m = 1 \ldots M$3 MHz subbands, under path-loss plus small-scale fading according to 3GPP UMi-Street-Canyon (Wilhelmi et al., 24 Jun 2026).

Traffic is modeled through Poisson arrivals at $m = 1 \ldots M$4 pkt/s per STA with packet size $m = 1 \ldots M$5 bytes. The reliability and latency targets are $m = 1 \ldots M$6 ms and reliability at least $m = 1 \ldots M$7, corresponding to frame loss probability at most $m = 1 \ldots M$8 (Wilhelmi et al., 24 Jun 2026). DSO control parameters include reconfiguration interval $m = 1 \ldots M$9 ms, tuned between $W_m$0 and $W_m$1 ms in sensitivity studies, CSI feedback periodicity of $W_m$2 ms, and weighting factors $W_m$3 chosen to prioritize delay-critical versus throughput-hungry flows (Wilhelmi et al., 24 Jun 2026).

These assumptions delimit the scope of the reported results. The paper’s configuration combines dense deployment, broad spectrum at $W_m$4 GHz, multi-antenna APs, and stringent UHR objectives; this suggests that the performance figures should be interpreted as representative of the stated operating regime rather than as universal constants for all Wi-Fi 8 deployments.

5. Comparative performance against legacy and static schemes

The comparison covers three schemes: legacy full-band operation, fixed static subbands with $W_m$5 MHz channels pre-assigned at boot, and Dynamic Subband Operation (Wilhelmi et al., 24 Jun 2026). Representative results are reported as means over $W_m$6 seeds and $W_m$7 s simulation time.

For aggregate throughput, the reported values are $W_m$8 Mbps for legacy, $W_m$9 Mbps for static subbands, and $\sum_m W_m = B_{tot}$0 Mbps for DSO. The static scheme is therefore reported as $\sum_m W_m = B_{tot}$1 over legacy, while DSO is reported as $\sum_m W_m = B_{tot}$2 over legacy and $\sum_m W_m = B_{tot}$3 over static (Wilhelmi et al., 24 Jun 2026). For $\sum_m W_m = B_{tot}$4th-percentile latency, the values are $\sum_m W_m = B_{tot}$5 ms for legacy, $\sum_m W_m = B_{tot}$6 ms for static, and $\sum_m W_m = B_{tot}$7 ms for DSO (Wilhelmi et al., 24 Jun 2026). For frame delivery ratio, the values are $\sum_m W_m = B_{tot}$8 for legacy, $\sum_m W_m = B_{tot}$9 for static, and $U$0 for DSO (Wilhelmi et al., 24 Jun 2026). Control overhead, counting beacon IE size plus CSI, is approximately $U$1 of airtime in DSO versus approximately $U$2 in static and legacy operation (Wilhelmi et al., 24 Jun 2026).

The accompanying interpretation is explicit: DSO’s ability to remap noisy or congested subbands away from delay-sensitive STAs yields the largest reliability and latency gains (Wilhelmi et al., 24 Jun 2026). Read together with the optimization model, these results indicate that the principal benefit of DSO is not only higher spectral efficiency but also more selective protection of deadline-sensitive traffic under heterogeneous channel conditions.

6. Design trade-offs, deployment guidance, and compatibility

The simulator study identifies several deployment recommendations. First, $U$3 MHz subband granularity is reported to strike the best throughput-reliability trade-off: finer splits at or below $U$4 MHz increase control overhead, while coarser splits at or above $U$5 MHz reduce adaptation gains (Wilhelmi et al., 24 Jun 2026). Second, a reconfiguration interval $U$6–$U$7 ms is stated to be sufficient to track medium-term fading and load changes, whereas faster adaptation yields only marginal further gains while increasing beacon size and computation (Wilhelmi et al., 24 Jun 2026).

Third, weight tuning is relevant for mixed URLLC and eMBB client populations: dynamic weight adaptation with larger $U$8 for backlog-heavy flows and larger $U$9 for deadline-critical flows is reported as beneficial (Wilhelmi et al., 24 Jun 2026). Fourth, in dense deployments, a lightweight inter-AP DSO coordination handshake via distributed conformance TWT is reported to avoid subband collisions and further improve spatial reuse (Wilhelmi et al., 24 Jun 2026). Fifth, backward compatibility is maintained by embedding DSO maps in standard beacon IEs so that legacy STAs experience no service interruption and maintain interoperability with non-DSO APs (Wilhelmi et al., 24 Jun 2026).

These points clarify the principal trade-off. DSO is associated with a modest increase in control overhead and more complex MAC scheduling, but the paper reports simultaneous gains in throughput, latency, and reliability under the evaluated conditions (Wilhelmi et al., 24 Jun 2026). A common misunderstanding is that compatibility requires separate legacy operation modes; the reported mechanism instead preserves coexistence by allowing legacy STAs to ignore the DSO IE and continue using the full band.

7. Interpretation and relation to broader Wi-Fi 8 research

Within Kom8ndor, DSO is one component of a broader 802.11bn feature set that also includes MAPC mechanisms such as Co-TDMA, Co-SR, and Co-BF, as well as NPCA and a machine learning wrapper for building AI-based protocols (Wilhelmi et al., 24 Jun 2026). In that broader architecture, DSO occupies the role of frequency-domain adaptation, whereas the other mechanisms address coordination, reuse, beamforming, and non-primary channel access.

The paper’s summary states that dynamically slicing and reassigning the spectrum every beacon allows Wi-Fi 8 APs to simultaneously boost throughput, slash latency, and achieve UHR targets, with only a modest increase in control overhead and somewhat more complex MAC scheduling (Wilhelmi et al., 24 Jun 2026). This suggests that DSO is best understood as a scheduler-level enabler for UHR rather than as an isolated PHY enhancement. Its value derives from the coupling of per-subband channel awareness, queue-awareness, deadline-awareness, and standards-compatible control dissemination.

In that sense, DSO in Kom8ndor provides a concrete operational model for how IEEE 802.11bn systems may exploit wide channels at $|U| = U$0 GHz without treating the band as a monolithic resource. The central idea is recurrent and explicit: the band is partitioned, evaluated, and reassigned at each control interval in response to current conditions, and the resulting gains emerge from that repeated adaptation (Wilhelmi et al., 24 Jun 2026).