An AP must have a unified, client-aware view of the network. That is, it must know how the network looks to its neighboring APs and to clients – both its own clients and those of the neighbors. The AP can then make informed steering decisions to ensure optimum client QoE.

Example: Minimum Association RSSI

To appreciate the value of a unified view of the network, consider the client in figure Minimum Association RSSI Example. It is located between two APs, AP1 and AP2. Suppose the client’s RSSI values, as seen by both APs, are lower than the minimum needed to associate with them. Then, without Unified Client Steering, the client cannot connect because neither AP1 nor AP2 accepts the client’s association request. With Unified Client Steering, however, AP1 is CV-CUE of the client’s RSSI as seen by AP2 and vice versa. Because AP1 knows that there is no neighboring AP that can see the client with an RSSI greater than the minimum association threshold, it does not reject the client’s association request. This allows the client to connect, improving user experience.

APs must not steer clients too frequently. Clients that are moving or happen to be in the coverage overlap region of two APs could “ping-pong” between the two APs because of constant back and forth steering. This is wasteful signaling and could cause poor user experience.

To avoid this, Arista APs should not attempt to steer a client too often. You can configure a Steering Attempts Threshold parameter that determines the maximum number of attempts to steer a client allowed in a 10-minute window (see Configuration section for details). The default value is 2. So, if an Arista AP has attempted to steer a client twice in 10 minutes, the client enters a configurable Blackout Interval (default 15 minutes). The AP does not attempt to steer such a client until the Blackout Interval has elapsed. An Arista AP shares the steering attempt epochs of its clients with its RF neighbors in its periodic wired-side broadcasts.

Example: Smart Steering

Figure Smart Steering Example shows a client located in the coverage overlap region between two APs, AP1 and AP2. The client’s RSSI could change quite frequently because of channel fading or because it might be moving. Without Unified Client Steering, when the client’s RSSI at AP1 drops below the configured threshold, AP1 steers it to AP2; when the RSSI at AP2 drops, the client is steered in the opposite direction. The client could thus constantly “ping-pong” between two APs. With Unified Client Steering, after being steered at most twice in 10 minutes, the client enters a 15-minute Blackout Interval (assuming all default values). This solves the client’s frequent “ping-pong” problem.

Multiple Input Multiple Output (MIMO) refers to the use of multiple antennas on Wi-Fi APs and clients to increase data rates (via spatial multiplexing) and reduce interference (via spatial diversity). Multiple antennas result in multiple, simultaneous “streams” of data between an AP and a client on the same channel. Multi-user MIMO (MU-MIMO) uses the multiple-antenna streams not to improve the transmission to a single user but to simultaneously serve multiple users, i.e., to improve capacity rather than user data rates. The figure below shows an example of MU-MIMO where four streams on the AP can simultaneously serve two clients, each with support for two streams.

Downlink MU-MIMO

Depending on the channel conditions of a client, an AP can use Single User MIMO (SU-MIMO) to improve the client’s data rates or reduce its error rates. Both, however, are per-link improvements. Since dense environments are about capacity and not just individual user data rates, downlink MU-MIMO can help in dense environments by allocating resources (i.e. streams) more efficiently among a large number of users. 802.11ax supports 8x8 MIMO on the downlink, i.e., an AP can simultaneously send data to eight users.

Uplink MU-MIMO

Uplink MU-MIMO is especially useful for uplink-heavy applications such as social media, content sharing, and video calls. As with the downlink, uplink MU-MIMO increases capacity compared to the SU-MIMO case. This results in a better user experience when uploading content. 802.11ax supports 8x8 MIMO on the uplink, i.e., eight users can simultaneously send their data to an AP. Early 802.11ax Wi-Fi clients might not support uplink MU-MIMO and for the ones that do, the implementation is not yet mature across client manufacturers. If your 802.11ax network has throughput problems, you might want to disable uplink MU-MIMO.

Orthogonal Frequency Division Multiple Access (OFDMA) divides the Wi-Fi channel into subcarriers, also called “tones”, each 78.125 KHz wide. Tones are combined to form Resource Units (RUs) of different widths. As shown in the representation below, an RU can consist of 26, 52, 106, or 242 tones—corresponding to channel widths of approximately 2 MHz, 4 MHz, 8 MHz, and 20 MHz. In each scheduling interval, OFDMA allocates one or more RUs of different widths to multiple users, resulting in an efficient and flexible use of the channel.

802.11ax uses OFDMA to support multiple users simultaneously, but the underlying multiple access mechanism continues to be CSMA-CA with backoff. Users are scheduled simultaneously, but only when CSMA-CA determines that the medium is available for transmission.

Downlink OFDMA

On the downlink, an 802.11ax Arista AP intelligently schedules clients and allocates resources. Until 802.11ac, an entire 20/40/80 MHz channel would be allocated to a single user during one time slot. This is not optimal because the client application may use only a fraction of the channel. Downlink OFDMA uses the channel much more efficiently by distributing it among clients, allocating only as much of the channel to each client as needed and changing the allocation when needed. The resource allocation and scheduling are based on factors such as the application QoS required and the channel conditions. An 802.11ax AP has the flexibility to allocate any combination of RUs. The figure below shows an example of an AP allocating RUs in three scheduling intervals.

In the first scheduling interval, the AP allocates two 52-tone RUs to Client 1 and Client 2, and one 26-tone RU to Client 3. In the second interval, it allocates the whole 20 MHz channel—a single, 242-tone RU—to Client 1. And in the third interval, it allocates two 106-tone RUs to Client 2 and Client 3.

Uplink OFDMA

The Wi-Fi uplink is a distributed form of communication because the transmitters (i.e., clients) cannot coordinate their schedules (unlike the downlink, where the AP is both the transmitter and the scheduler). Until 802.11ac, Wi-Fi uplink was uncoordinated: clients contended for the medium and, based on randomly distributed timing, sent packets to the AP. This works reasonably well for single or sparse AP deployments, but for dense deployments, this can cause high uplink contention in presence of a large number of clients. With 802.11ax, the Wi-Fi uplink is coordinated. The AP manages the uplink resource allocation via mechanisms that essentially coordinate the transmission schedule among clients while using CSMA-CA to ensure that the medium is available for transmission. This leads to reduced uplink contention, thereby improving capacity and user experience. Early 802.11ax Wi-Fi clients might not support uplink OFDMA and for the ones that do, the implementation is not yet mature across client manufacturers. This may adversely affect the 802.11ax AP throughput. If your 802.11ax network has throughput problems, you might want to disable uplink OFDMA.

With spatial reuse, two or more Wi-Fi devices (AP or client) that support 802.11ax protocols can send transmissions simultaneously without any significant data loss. Spatial Reuse helps in improving the spectral efficiency and optimal allocation of resources to meet the Quality of Service (QoS).

How Spatial Reuse Improves Efficiency?

To understand spatial reuse, consider the two co-channel BSSs as shown in figure below.

BSS1 and BSS2 operate on the same channel and can hear each other with an RSSI higher than the Clear Channel Access Signal Detection (CCA-SD) level, the threshold that Wi-Fi devices use to decide if the channel is clear to transmit or if they need to wait for the . A co-channel BSS that can be heard at an RSSI greater than the threshold is called an overlapping BSS (OBSS). Thus, BSS2 is an OBSS for BSS1, and vice versa.

Before Wi-Fi 6, if a Wi-Fi device detected an ongoing transmission with RSSI higher than CCA-SD on a specific channel, all other devices on the same channel had to back off and they would wait for the ongoing transmission to get over. With Wi-Fi 6, a device can leverage spatial reuse to transmit in parallel with an ongoing OBSS transmission, provided the spatial reuse transmission is at an acceptably low transmit power.

BSS Color

A Wi-Fi device that wants to transmit using spatial reuse needs to distinguish the transmissions of its own BSS from those of the OBSS. To enable this, Wi-Fi 6 introduces BSS color, an integer between 1 and 63 to identify a BSS. An AP radio automatically assigns a BSS color to each transmission. When two radios transmit on the same channel, BSS color helps an AP to differentiate between an inter-BSS transmission and intra-BSS transmission.

Typically, every 802.11ax HE (High Efficiency) frame contains a BSS color value that a Wi-Fi device can use to identify the frame as its own BSS or an OBSS. The BSS color is signaled in the 802.11ax PHY preamble, allowing for early detection of an OBSS transmission. Since a Wi-Fi 6 device instantly decodes the HE preamble, the device can identify an OBSS transmission well in advance to transmit in parallel.

As long as BSSs in vicinity of each other use distinct BSS colors, a device can distinguish the transmissions of its own BSS from those of an OBSS.

Enabling Spatial Reuse