In the past, IEEE 802.11 standards (g, a, n, ac) delivered WiFi performance improvements out of the box. They focused on progressively increasing the data rate over the wireless link. All it took to take advantage of any new standard was a radio chipset that incorporated the new radio and MAC enhancements. Also, since chipsets become available concurrently with the standard, there wasn't a time lag between the standard ratification and realization of its benefits in practice.
The situation is different for the upcoming 802.11ax standard. The focus of 802.11ax is not on increasing the data rate, but on improving the overall wireless network performance. To this end, it introduces significant new radio and MAC enhancements such as OFDMA and BSS coloring. However, intelligent mechanisms on the AP are required to derive these benefits. The rub is that such mechanisms are yet to be devised and tested in practice.
Ranking high among the issues is a transmission-scheduling mechanism. The downlink transmission scheduling in WiFi has been a simple FIFO (First In First Out) system. 802.11e introduced a small variation regarding the maintenance of multiple transmission queues for different priority classes. 802.11n and 11ac brought frame aggregation, but it was mostly done ad hoc without much intelligence applied to it. Also, the AP used to transmit to only one client at a time. In the reverse direction, only one client used to transmit to the AP at a time.
However, 802.11ax introduces significant complexity in wireless transmission scheduling due to its OFDMA and MU-MIMO enhancements. MU-MIMO is technically an .11ac Wave 2 feature, but given the scant use of MU-MIMO in practical deployments so far, it's appropriate to discuss it in the context of 802.11ax. An access point that supports 802.11ax needs to perform complex decision making for transmission scheduling. Following are some of the considerations, with much devil in the detail of each:
- With MU-MIMO, there is now an option to transmit a single wireless frame to a single client or concurrently transmit different wireless frames to multiple clients using multi-user beamforming.
- With OFDMA, there is now an option to transmit a single wireless frame to single client using traditional OFDM or concurrently transmit different wireless frames to multiple clients using subsets of channel width.
- 802.11ax introduces multi-user transmission in uplink direction too. The AP needs to schedule multiple clients for concurrent uplink transmissions according to their requirements.
These methods need to take into account service requirements of traffic flows, radio conditions on the channel, client capabilities, and client state feedbacks. It is no easy feat to come up with scheduling mechanisms that will work in most practical scenarios with relative ease of configuration and fine tuning.
The other significant 802.11ax MAC feature, BSS coloring, requires dynamic calibration of carrier sense threshold in the BSS, which has been a static threshold in all WiFi standards so far. The BSS coloring is designed to increase spatial frequency reuse in WiFi in order to raise overall network performance. However, the actual benefit in practical deployments will depend on the ability of APs and clients to converge to the optimal carrier sense threshold in their BSS. Estimating this parameter correctly is a complex task and a wrong choice can cause performance hit rather than gain.
802.11ax started its journey at the beginning of the hype cycle one year ago. Due to the upcoming ratification of the standard and the announced availability of 802.11ax chipsets by the leading vendors, one would expect it to move upwards on the hype cycle to the “peak of expectations.” However, due to the lack of field-tested mechanisms as described above, I think we will see 802.11ax making its way through the “trough of disillusionment” in the near future.
In my last blog, I wrote about promise of 802.11ax for IoT. It was a secondary consideration in the design of 802.11ax with network access performance being the primary. IoT applications are not performance hungry. So, even though we have to wait and watch for 802.11ax to deliver on network access performance in practice, the IoT use case could still propel 802.11ax adoption.