There has been some consternation that 802.11ax should have a greenfield mode in 6 GHz, leaving behind all the protocol overhead used for backwards compatibility in the 2.4 and 5 GHz bands. This mythical mode could also have fantastic new capabilities that would now be possible without legacy PHY requirements. 6 GHz is an opportunity to so radically overhaul 802.11 that we could increment the 802.11 version bit in all 802.11 6 GHz frames (It’s been 0 for the entire history of Wi-Fi). Of course, it probably isn’t reasonable to expect the same amendment that must provide backwards compatibility in the legacy bands to also do something radically new in 6 GHz. It may also be unrealistic to expect 802.11ax client chipsets that operate in the legacy bands in legacy modes to do something radically different using the same radio in 6 GHz. Still, there are real protocol differences in 6 GHz 802.11ax operation. 802.11ax is not ratified, so it is still possible for some things to change, but I thought I’d run down what’s different and what opportunities I think were missed with 802.11ax in 6 GHz.
- Security Upgrade – 802.11ax will make SAE and OWE mandatory replacements for PSK and open auth respectively in 6 GHz. MFP will be required. I still don’t understand how this will work with SSID’s that span the legacy bands to support legacy WPA2 clients as well as WPA3-only clients in 6 GHz. If the answer is “Just add another SSID for Wi-Fi 6E clients-only,” then I will be disappointed.
- OFDM and HE-Only – There is no HT or VHT operation allowed in 6 GHz. Unique HE beacon IE’s indicate support for features inherited from HT and VHT. There are no HT/VHT Capabilities IE’s in a 6 GHz beacon. No HT or VHT MCS will be used in 6 GHz. OFDM is there because its shorter preamble consumes less airtime than the new HE preamble, so it will be used for those frames that don’t require the bigger HE preamble.
- Basic HE-MCS and NSS (number of spatial streams) Set – We aren’t using legacy rates outside of the legacy preamble, RTS/CTS, and legacy ACK frames. Most frames can be modulated with HE-MCS encoding, including beacon, multi-STA blockACK, and trigger frames, some of which can be transmitted with multiple spatial streams.
- Spatial Reuse Can Work – 6 GHz STA’s can take full advantage of BSS Coloring and OBSS CCA-PD. Without legacy STA’s to conflict with, we can design for these features to significantly desensitize all intra-BSS STA’s from OBSS frames, and allow for increased spectral efficiency by reusing the channel more aggressively. If implemented, those robustly-modulated preambles are a smaller problem. However, Spatial Reuse (the OBSS CCA-PD) is optional in 802.11ax. Dual NAV is required for clients, and optional for AP’s. Confused yet?
- EDCA Optimization – This might be the trick to getting OFDMA operation to take place more often. All 6 GHz STA’s will support OFDMA, so why not increase the contention window for the SU EDCA access categories advertised in beacons (there is a separate MU EDCA table for OFDMA)? That would reduce the likelihood of clients winning access to the channel for SU operation, and increase the likelihood that the AP will win the channel for OFDMA operation to take place.
- Less RTS/CTS Overhead? – Because all 6 GHz STA’s can interpret the HE preamble, which includes the duration of the TxOP, normal RTS/CTS protection is redundant. The 802.11ax draft allows for several ways to establish a TxOP, including the legacy duplicate RTS/CTS method, so we will have to wait and see what the vendors choose to use in 6 GHz. In ideal circumstances, 802.11ax in 6 GHz will look like this: AP wins arbitration, trigger frame, MU-PPDU, BlockACK, repeat, repeat, repeat…
- Reason Code 71 – 6 GHz AP’s can deny an association request from a client with poor RSSI using status “DENIED_POOR_CHANNEL_CONDITIONS” or disassociate a low RSSI client with a new reason code 71,”POOR_RSSI_CONDITIONS.” A 6 GHz client must respond to this sensibly (e.g. not blacklisting the BSSID/SSID as clients sometimes do in the legacy bands). Although vendors have had features that accomplished this for a long time, client behavior in response has always been a mixed bag.
- 6 GHz AP Discovery and Association – A 6 GHz STA can discover, and in some cases associate to, a 6 GHz radio while operating in the 2.4 or 5 GHz bands. An AP’s beacon, probe response, and neighbor report frames in those bands can indicate the channel and channel width of their matching 6 GHz radio. Additionally, for 6 GHz-only operation, a specific subset of channels will be identified as preferred scanning channels (PSC) where the primary channel of a wide channel BSS should reside, limiting the channels a client needs to scan to discover a 6 GHz-only AP. PSC’s are spaced 80 MHz apart, so a client would only need to scan 14 channels in the US. Active probing in the 6 GHz band in the US is only allowed after a client has heard an AP transmission on the channel, which includes a beacon frame, an unsolicited probe response sent to the broadcast address, or a FILS discovery frame. However, one of these frames can be transmitted at least every 20 TU’s, which allows for less required dwell time on the channel for passive scanning. Less required dwell time and a limited set of PSC’s will make passive AP discovery faster in 6 GHz than 5 GHz.
- 80 MHz AP Channel Width Minimum in 6 GHz – They weren’t lying when they said “80 is the new 20.” Maybe things will change before 802.11ax is ratified, but I’ve learned that a 6 GHz AP will have to indicate support for at least 80 MHz channel width. This aligns well with the PSC’s the standard will define. I don’t know why the IEEE would require this, as it is extremely undesirable in the LPV WLAN’s where 802.11ax in 6 GHz would otherwise provide the most benefit. It’s an even larger problem in countries with unlicensed access to a smaller portion of the band. Clients may still use smaller channel widths, including a 20 MHz-only operating mode.
- How Much of the Wide Channel Can Be Used? – The use of wide channels followed wasteful logic in 802.11ac with dynamic bandwidth operation (DBO): If the primary 20 MHz channel of a 160 MHz BSS is busy nothing can be transmitted. If the secondary 20 MHz channel that would be used for a 40 MHz channel-width is busy then the STA can only use 20 MHz, the rest of the 120 MHz is unused. If those first two 20 MHz channels are clear, another 40 MHz is checked to see if 80 MHz of the channel is available, etc. And this pattern of checking for the next wider channel width is done serially through the CCA process which also introduces more overhead on its own. Remember, OFDMA only happens within the TxOP gained from a wide-channel arbitration process, which can be subject to the logic I just described. DBO was optional in 802.11ac and not widely implemented, and it appears to be optional in 802.11ax as well. This is important because…
- Preamble Puncturing is Optional – What improves spectral efficiency in the scenarios above is a new 802.11ax feature called preamble puncturing, which allows a HE STA to transmit across the full 160 MHz of spectrum, but not within the specific 20 MHz subchannels that are busy. One busy subchannel doesn’t prevent the use of others, but the primary channel must still always be free for anything to be transmitted. However, preamble puncturing is an optional feature in 802.11ax. So even in the best case scenarios, OFDMA can only subdivide the channel within the 20 MHz subchannels determined to be available via CCA and (maybe) RTS/CTS. In the worst case scenario (no DBO or preamble puncturing), no data frame, OFDMA or otherwise, can be transmitted unless the entire wide channel is available (minimum 80 MHz in 6 GHz!).
The primary channel bottleneck is particularly troublesome because future generations of 802.11 in 6 GHz will have to account for 802.11ax operation to maintain backwards compatibility, although a lot of traffic that used to be primary channel-only can be included in OFDMA transmissions now (ACK, null-data frames). I wish we could have left behind the legacy channel arbitration process as well, or at least made preamble puncturing mandatory. Other problems may be improved in future amendments. The potential for reduced overhead and higher likelihood of the AP winning channel access are significant improvements when coupled with all 802.11ax clients in 6 GHz.
To Be Determined
- The Wi-Fi Alliance – The WFA could decide that features that are optional in 802.11ax are mandatory for Wi-Fi 6E certification. Preamble puncturing and spatial reuse are very beneficial features that should be mandatory.
- 802.11md – This is the maintenance work being done to roll-up 802.11 into 802.11-2020. It will include 802.11ax-2020 and leaves open the possibility of additional changes to 6 GHz operation occurring that are not part of the 802.11ax drafts, but happen within the roll-up process separately. An example of this happening in the past is 802.11 Fine Timing Measurement, which was added to the standard through the 802.11mc roll-up to 802.11-2016. That feature does not have its own 802.11xx amendment. Hopefully this will all be sorted out in December. Stay tuned.
- 802.11ax, TBH – It’s not finalized yet, so perhaps some of this will change. I’ll update this blog if that happens.