All posts
Wi-Fi · MAC

Non-Primary Channel Access (NPCA) in Wi-Fi 8

By Gopi Raga · June 2026

In Wi‑Fi, channel access lives or dies on the primary 20 MHz channel. If a neighbor occupies it, even a 320 MHz channel can freeze while every other subchannel sits idle. Wi‑Fi 8 fixes this with Non-Primary Channel Access: when the primary is blocked, the station relocates to clean spectrum it already owns and keeps working.

The problem

Before a device can transmit, it senses the primary 20 MHz channel. If that channel is busy, the device has to defer; it cannot transmit until the channel goes idle.

The painful part is what happens to the rest of your bandwidth. If a neighboring Overlapping BSS (OBSS) access point is transmitting on your primary 20 MHz, your entire channel width is blocked along with it. Even if you have configured a wide channel such as 80, 160 or 320 MHz, and every other subchannel is sitting completely idle, the whole channel is frozen simply because the primary 20 MHz is busy. All of that spectrum goes to waste.

The fix

Wi‑Fi 8 introduces Non-Primary Channel Access (NPCA), a mechanism that lets a station dynamically switch operation from the BSS primary channel to an alternate primary channel within the same BSS when the primary 20 MHz is found busy due to OBSS transmissions. Instead of sitting idle while a neighbor occupies the primary, the device relocates to clean spectrum it already owns and keeps working, turning blocked airtime into usable capacity.

Two prerequisites must be satisfied before NPCA can be used:

  • Both the AP and the client must support the feature and advertise it during association.
  • The AP's operating bandwidth must be 80 MHz or wider. The feature works by moving to a spare subchannel, and those spare subchannels only exist on a sufficiently wide channel.

Step 1: The AP enables NPCA and the client negotiates

Which channel does the AP pick? It cannot be random. The alternate channel must come from the secondary half of the AP's bandwidth. For example, if the AP is configured for 160 MHz, it selects any 20 MHz channel from within the secondary 80 MHz. This guarantees the alternate channel is always separate from the primary being interfered with, since the primary always lives in the primary half.

How long can the device stay on the new channel? This is governed by the NPCA Minimum Duration Threshold, which has a minimum value of 512 µs (and a maximum of 2432 µs). The 512 µs floor is deliberately chosen to be long enough to cover at least one useful exchange on the new channel: a control frame handshake followed by a data frame and its acknowledgment.

The AP advertises these values, along with the NPCA Switching Delay and NPCA Switching Back Delay, in its Beacon and Probe Response frames. A client that decides to participate then sends a UHR OMP request to the AP asking to enable NPCA mode, and in that request it reports its own NPCA Switching Delay and NPCA Switching Back Delay so the AP knows how quickly it can move.

It is worth noting that NPCA is disabled by default when a client first associates. Enabling it is an explicit, per-client negotiation, and even once enabled, NPCA stays dormant until a triggering condition occurs.

Step 2: The trigger, when a switch actually happens

A switch to the alternate primary channel is allowed only if NPCA is enabled and all of the following conditions are met:

  • The busy transmission must be classified as inter-BSS, meaning it genuinely comes from a neighbor and not from the device's own network.
  • The offending transmission's channel must not overlap the alternate primary channel, otherwise the destination would also be occupied.
  • If the station holds an intra-BSS NAV, it must be zero, confirming the device is not in the middle of its own exchange.
  • The estimated remaining busy time must exceed the Minimum Duration Threshold of 512 µs, after accounting for the switching overhead.

On top of these common conditions, the device must confirm the situation through one of two scenarios.

Scenario 1: a single long neighbor transmission

The device detects a long inter-BSS PPDU on the primary channel and determines its remaining airtime from the PPDU's PHY header at the start of reception, stored as NPCA_PPDU_REM_DUR.

For example, suppose a neighbor's PPDU has 800 µs remaining, and assume this device has a switching delay and switch-back delay of 20 µs each (the standard expresses these in units of 4 µs; 20 µs is used here only as an example):

800 − 20 − 20 = 760 µs   >   512 µs  →  switch ✓

If the PPDU had only 400 µs left:

400 − 40 = 360 µs   <   512 µs  →  do not switch ✗

The detour simply would not be worth the time it takes to make it.

Scenario 2: a neighbor has won an entire TXOP

Rather than reacting to a single frame, the device recognizes that a neighbor has secured a full transmit opportunity (TXOP). It does this by detecting the sequence of frames that establishes a TXOP on the primary channel, an initial control frame followed by its response and then the data transmission that follows, and confirming that a valid TXOP was obtained. It then estimates how much of that TXOP remains and applies the same threshold test.

This whole-TXOP estimate is available when the AP has enabled the MOPLEN mode of NPCA, which lets the device use duration information from the MAC and PHY headers, for example the duration value carried in the control frame that opened the TXOP. If only the simpler PHYLEN mode is enabled, the device relies on per-frame PHY duration instead. Either way, the advantage of Scenario 2 is a more accurate view of how long the primary channel will stay occupied.

Step 3: The switch to the alternate channel

Once the decision is made, two timing rules keep both ends synchronized:

  • The client must be ready to transmit and receive on the alternate primary channel no later than its advertised switching delay.
  • Mutual-arrival guarantee: a transmitter must not send to a peer until that peer's switching delay has elapsed. This ensures both devices have completed the move before any traffic is exchanged, so nothing is sent to a device that has not yet arrived.

Step 4: Operating on the alternate channel

When the device arrives on the new channel, an NPCA_TIMER begins counting down in 1 µs steps. Its initial value equals the largest valid remaining-duration estimate minus the switch-back delay. In other words, it represents how long the device is allowed to stay on the alternate channel before it must head back, and it is set so the device returns to the primary before the neighbor's transmission ends.

Several rules govern behavior while on the alternate channel:

  • Channel occupancy: any transmission must include at least the alternate primary channel, stay within the BSS bandwidth, and must not overlap the OBSS channel that triggered the switch.
  • Robust framing: each frame exchange begins with a control frame sent at a low, widely decodable data rate, so surrounding devices correctly interpret the activity.
  • Restricted operations: the device does not send Probe Request, Probe Response or FILS Discovery frames on the alternate channel.
  • Beacons deferred: if the AP's beacon time (TBTT) falls while it is operating on the alternate channel, the Beacon and any group-addressed frames are postponed until switch-back, buffered and delivered after the next DTIM beacon.

Step 5: Switching back to the primary channel

When the NPCA_TIMER expires, the client switches back to the BSS primary channel and restores the channel-access state it had saved before leaving, resuming as though nothing had happened. By design, the device arrives back on the primary at roughly the moment the neighbor finishes, so it is immediately ready to contend for the primary again.

— Gopi Raga