802.11ac Wave 2 is rising fast, as legacy 802.11n networks are upgraded. Current 802.11ac Wave 1 networks may not upgrade to Wave 2 right now, but by the time these networks have hit their typical 4-5 year lifespan, the next WiFi standard will be available: 802.11ax.
From raw speed to efficiency
The IEEE 802.11 evolution so far (b, g, a, n, ac) has delivered increasingly higher data rates with better modulation, channel bonding, MIMO, etc. However, nuances of real-life WiFi usage, particularly in heavy usage or dense scenarios, prevent the increased data rates from fully benefiting the end user. The upcoming 802.11ax standard bucks this trend by focusing on efficiency instead of raw data rate.
802.11ax vs. LTE-U/LAA
There is a possibility of LTE-U/LAA becoming the dominant radio technology in the unlicensed spectrum for heavy or dense usage scenarios. It is generally considered more efficient than WiFi, because of the LTE lineage. The flip side is that LTE-U/LAA requires control carrier in the licensed LTE spectrum. LTE-U/LAA is preferred by telecom carriers who own LTE licensed spectrum, but is not suitable for players who don’t. By incorporating new efficiency measures in the WiFi protocol, 802.11ax could become the viable pure play unlicensed alternative to LTE-U/LAA.
Antidote to small frames
Empirical evidence shows the prevalence of small sized frames in real-life WiFi traffic. In one published data set, as many as 80% of the frames in WiFi traffic are found to be small (under 256 bytes). Small frames can be of data, management or control types. Each small frame incurs fixed overhead of CSMA/CA access, preamble, and (for data frame) ACK. Even if ACK is transmitted without CSMA/CA, it still has a preamble that is large compared to its payload: at a 12 Mbps data rate, the preamble is 20 microseconds and the BACK/ACK payload is 20/12 microseconds. To address the problem of small frames reducing the usable capacity of the radio channel, 802.11ax uses OFDMA.
802.11ax introduces Orthogonal Frequency-Division Multiple Access, a backward-compatible enhancement to OFDM. In OFDM, the total channel bandwidth (20 MHz, 40 MHz, 80 MHz, etc.) contains multiple OFDM sub-carriers. Presently in OFDM, any one frame transmission has to use all the sub-carriers in the bandwidth. However, in OFDMA, different subsets of sub-carriers in the channel bandwidth can be used by different frame transmissions at the same time. (This is not to be confused with MU-MIMO of 802.11ac Wave 2.) Sub-carriers can be allocated to transmission in blocks as small as 2 MHz.
With OFDMA, 802.11ax allows parallel transmissions of small frames to different receivers. Thus, the fixed overhead is incurred only once for all parallel frames. In one of the contributions to the standards committee, it has been stated that the time to transmit four frames of 90 octets each and their ACKs sequentially using conventional OFDM on 80 MHz channel is more than 400 microseconds, while the time to transmit them in parallel using OFDMA in both directions is under 150 microseconds. That is about 3 times increase in the capacity of channel as it pertains to transmission of small frames.
Astute readers must have noticed that OFDMA in the downlink is relatively easy to maneuver because the AP can schedule the frames in parallel, but how does OFDMA work in the uplink, where different transmissions are coming from distributed stations? That is where uplink scheduling comes in.
In 802.11ax, the AP can provide uplink transmission schedules to clients. In situations such as ACKs, this is easy as they immediately follow data packets. However, for other situations some information feedback from the clients may be required for the AP scheduler to estimate clients’ transmission requirements. Feedback can come piggybacked on previous frames from clients or via explicit notifications.
The AP provides transmission schedules to clients using a new control frame called a “trigger frame,” which specifies which clients can transmit during the specified time and which subsets of OFDMA sub-carriers they will use. Clients can also use scheduling algorithms to determine which frames they need to transmit immediately using conventional OFDM, and which ones to keep waiting for allocation of OFDMA schedules.
The caveat here is that OFDMA can only be as good as these scheduling algorithms. Vendors need to come up with intelligent scheduling to leverage OFDMA efficiently. Fingers crossed!
There’s more in 802.11ax
There are other notable enhancements in 802.11ax: dynamic sensitivity control and support for long range transmissions. I will cover them in another blog post. Stay tuned!