WLAN Transmit Beamforming (TxBF)

1. Definitions, acronyms, and abbreviations

• beamformee: A station (STA) that receives a physical layer (PHY) protocol data unit (PPDU) that was transmitted using a beamforming steering matrix.
• beamformer: A station (STA) that transmits a physical layer (PHY) protocol data unit (PPDU) using a beamforming steering matrix.
• beamforming: A spatial filtering mechanism used at a transmitter to improve the received signal power or signal-to-noise ratio (SNR) at an intended receiver. Syn: beam steering.
• beamforming steering matrix: A matrix determined using knowledge of the channel between a transmitter and an intended receiver that maps from space-time streams to transmit antennas with the goal of improving the signal power or signal-to-noise ratio (SNR) at the intended receiver.
• high-throughput (HT) null data PPDU (NDP) announcement: A physical layer (PHY) protocol data unit(PPDU) that contains one or more +HTC frames (i.e., frames with an HT Control field) that have the HT NDP Announcement subfield equal to 1.
• multi-user (MU) beamformee: A non-access-point (non-AP) station (STA) that receives a physical layer(PHY) protocol data unit (PPDU) that was transmitted using a multi-user beamforming steering matrix and that supports the very high throughput (VHT) transmit beamforming feedback mechanism with a VHT null data PPDU (NDP) Announcement frame that includes more than one STA Info field.
• null data PPDU (NDP): A physical layer (PHY) protocol data unit (PPDU) that carries no Data field.
• high-throughput (HT) beamformee: An HT station (STA) that receives an HT physical layer (PHY)protocol data unit (PPDU) that was transmitted using a beamforming steering matrix and that supports an HT transmit beamforming feedback mechanism.
• high-throughput (HT) beamformer: An HT station (STA) that transmits an HT physical layer (PHY)protocol data unit (PPDU) using a beamforming steering matrix.
• very high throughput (VHT) beamformee: A VHT station (STA) that receives a VHT physical layer(PHY) protocol data unit (PPDU) that was transmitted using a beamforming steering matrix and thatsupports the VHT transmit beamforming feedback mechanism.
• very high throughput (VHT) beamformer: A VHT station (STA) that transmits a VHT physical layer(PHY) protocol data unit (PPDU) using a beamforming steering matrix.

2. TxBF Overview

波束成形技术是将信号以一种能量集中和定向方式发送给无线终端的技术，能全面改善无线终端接收的信号质量，并提升吞吐量。在Wi-Fi标准中，从Wi-Fi 4(802.11n)开始引入该技术。

Transmit Beamforming (TxBF) is the use of multiple antennas to transmit a signal strategically with varying phases so that the communication arrives at the receiver in such a manner that the signal strength is increased. Since transmit beamforming results in constructive multipath communication, the result is a higher signal-to-noise ratio and greater received amplitude. The use of transmit beamforming results in greater range for individual clients when communicating with an AP. Transmit beamforming allows for higher throughput due to the higher SNR. This allows for the use of more-complex modulation methods with the ability to encode more data bits. The higher SNR also allows communications with fewer Layer 2 retransmissions, further improving the efficiency.

All this improvement in communication requires more than wishful thinking on the part of the transmitting STA. This requires communications between the transmitter and receiver about the quality of the signal received and how to improve it for subsequent transmissions. This feedback comes in two forms, Implicit and Explicit methods of calculation.

For an HT or VHT transmitter, called a beamformer, to calculate an appropriate steering matrix for transmit spatial processing when transmitting to a specific HT or VHT receiver, called the beamformee, the beamformer needs to have an accurate estimate of the channel over which it is transmitting. The transmitter knows what it hears but needs to learn from the receiver what it hears to better form the next transmission. When using the implicit feedback method, the beamformer receives long training symbols transmitted by the beamformee, which allow the MIMO channel between the beamformee and beamformer to be estimated. When the MIMO channel being used is reciprocal, the beamformer can use the training symbols that it receives from the beamformee to make a channel estimate which is suitable for computing the transmit steering matrix to be used in the next transmission.

Calibrated radios in MIMO systems can improve reciprocity. The HT TxBF calibration procedure consists of 4 frames.

• Calibration Start frame / Position 1 / sent by calibration initiator / TRO=1: This Sounding PPDU (A QoS Null Data frame) initiates the calibration procedure and is followed by an ACK frame.
• Calibration Sounding Response frame / Position 2 / sent by calibration responder: This Sounding PPDU (an ACK+HTC frame) is sent a SIFS after receipt of the Calibration Start frame. This frame is used by the calibration initiator to estimate the MIMO channel.
• Calibration Sounding Complete frame / Position 3 / sent by calibration initiator: This Sounding PPDU is sent a SIFS after receipt of the Calibration Sounding Response frame. This frame is a OoS Null+HTC frame, has the CSI/Steering subfield of the HT Control field set to 1, is used by the calibration responder toestimate the MIMO channel.
• A normal ACK frame (the 4th frame in the calibration sequence)

Each calibration sequence has a unique identifier found in the Calibration Sequence subfield in the HT Control field. This identifier remains the same throughout each calibration sequence and is incremented each timea new calibration procedure is started. There are specific requirements for both the beamformer and beamformee when using implicit feedback. The procedures for HT transmit beamforming with implicit feedback use only HT and non-HT PPDUs. The HT Control field, when present, is the HT variant HT Control field.

Transmit beamforming with implicit feedbas the ability to operate in aunidirectional or bidirectional manner. When using unidirectional implicit transmit beamforming, only the HT beamformer sends beamformed transmissions. When using bidirectiona limplicit transmit beamforming, both STAs send beamformed transmissions. This means that a STA can act as both an HT beamformer and an HT beamformee. Using implicit feedback, the calibration of receive or transmit chains should be done to improve performance of transmit beamforming. Over-the-air calibration is described in 10.32.2.4 of 802.11-2016. For implicit transmit beamforming, only the HT beamformer, which is sending the beamformed transmissions, needs to be calibrated. A device that advertises itself as being capable of being either an HT beamformer, an HT beamformee using implicit feedback, or both will be required to operate as described below.

Tha beamformer must:

• Set the Implicit Transmit Beamforming Capable subfield to 1 of the
• Transmit Beamforming Capability field of the HT Capabilities element in HT
• Capabilities elements that it transmits.
• Set the Implicit Transmit Beamforming Receiving Capable subfield to 1
• of the Transmit Beamforming Capability field of the HT Capabilities element.
• Be capable of receiving a sounding PPDU for which the SOUNDING
• parameter is SOUNDING and the NUM_EXTEN_SS is equal to 0 in the
• RXVECTOR in the PHY-RXSTART indication primitive, independently ofthe values of the Receive Staggered Sounding Capable and Receive NDP
• Capable subfields.
• Set the Calibration subfield to 3 of the Transmit Beamforming Capabilityfield of the HT Capabilities element to advertise full calibration support.

The beamformee must:

• Shall set the Implicit Transmit Beamforming Receiving Capable subfield to 1 ofthe Transmit Beamforming Capability field of the HT Capabilities element in HTCapabilities elements that it transmits.
• Shall be capable of setting the SOUNDING parameter to SOUNDING and the NUM_EXTEN_SS to 0 in the TXVECTOR in the PHY-TXSTART request primitive when transmitting a sounding PPDU, as a response to TRQ=1, independently of the values of the Transmit Staggered Sounding Capable and Transmit NDP Capable subfields.

Any STA that performs one of the roles related to transmit beamforming with implicit feedback must support the associated capabilities you can see in following figure. To assist the transmitter and receiver in transmit beamforming the devices use sounding PPDUs which are used to help determine the direction in which the signal is to be steered. It is similar to a Wi-Fi version of the swimming game Marco Polo, where one person says Marco, to ask where you are and the other says Polo, to be found.

In many cases it is desirable to obtain as full a characterization of the MIMO channel as is possible. To do this the transmission of a sufficient number of High Throughput Long Training Fields, HT-LTFs, is required to sound the full dimensionality of the channel. As stated in the 802.11 standard, if the HT beamformee transmits a sounding PPDU, the SOUNDING parameter in the TXVECTOR in the PHY-TXSTART request primitive shall be set to SOUNDING. If the HT beamformee is capable of implicit transmit beamforming and the HT beamformer can receive implicit transmit beamforming, the sounding PPDU from the HT beamformee may be steered. A STA that acts as an HT beamformer using implicit feedback expects to receive a sounding PPDU from the beamformee in response to a training request. The beamforming STA can then compute steering matrices from the channel estimates obtained from the sounding PPDU received from the beamformee.

Implicit transmit beamforming can be either unidirectional or bidirectional. When using unidirectional TxBF, the 802.11 standard states that the PPDU exchange can besummarized as follows:

• STA A initiates the frame exchange sequence by sending an unsteered PPDU to STA B. The PPDU includes a training request (TRQ=1) in a +HTC MPDU.
• STA B sends a sounding PPDU in response to the training request from STA A.
• On receiving the sounding PPDU, STA A uses the resulting channel estimate to compute steering matrices and uses these matrices to send a steered PPDU back to STA B.
• The steered PPDU transmitted in step c) and subsequent steered PPDU stransmitted by STA A may include training requests (TRO=1) in a +HTC MPDU. In response to each training request, STA B returns sounding PPDU to STA A, which enables STA A to update its steering vectors. If the steering vectors resulting from step c) or subsequent sounding PPDUs are deemed stale due todelay, the sequence may be restarted by returning to step a). Step d) in the above PPDU exchange represents steady-state unidirectional transmit beamforming operation.

During the PPDU exchange neither the receiving nor the transmitting STA should switch antennas.

2.1 TxBF Block Diagram

2.1.1 Wi-Fi为什么要用波束成形？

Wi-Fi标准一直致力于提升无线的传输速率，尤其是从Wi-Fi 4(802.11n)开始引入了MIMO和波束成形技术，让传输速率提升到了数百兆，提升了1个量级。

MIMO技术通过多天线传输，带来传输速率的成倍增长。但在实际应用中STA(无线终端)往往只有1到2个天线，这使得STA发送和接收信号的收益有所差异。STA向AP发送信号时，AP可以利用自己的多天线系统增强接收增益，获得更好的信号强度；AP向STA发送信号时，如果仅使用对应数量的天线发送信号，则无法利用多天线带来的增益。为了解决这一问题，通过引入波束成形技术，可以增强STA接收到的信号强度，从而使AP和STA可以协商出更高的传输速率。

为了充分利用AP的多天线资源，Wi-Fi 5(802.11ac)又引入了MU-MIMO技术，使AP可以同时向多个STA发送信号，有效提升了无线的传输效率。MU-MIMO也需要波束成形技术，波束成形使AP的多天线信号叠加后，让各STA仅收到自己的信号，消除其他STA的信号，避免干扰。Wi-Fi 6(802.11ax)在Wi-Fi 5的技术上进一步增加了MU-MIMO的多用户数量，这些都离不开波束成形技术的使用。

2.1.2 波束成形是如何工作的？

波束成形从字面理解就是塑造波束的形状，那么如何塑造出波束的形状呢？以光束为例，用一个手电筒打出一道光束，光束的形状是固定的。如果在平行方向增加一个手电筒，则会发现两道光束叠加后，光束亮度增加，光束形状发生了改变。如果继续增加手电筒的数量，光束的亮度继续增加，叠加后的光束形状也继续变化。在多个手电筒的情况下，改变手电筒的开关状态或调整发光的强弱，也会影响光束的形状。

对于无线通信，天线就相当于手电筒，无线信号的波束就相当于光束，通过多个天线，控制每个天线发射的信号，就可以改变无线信号的波束形状。

在多天线系统中，如果不同天线传输的信号在到达某一位置时存在两条衰减相等的波束，且两条波束相位相反，就可能会出现空间空洞。波束成形技术可以通过预先补偿发射天线的相位，让两条波束进行叠加以实现信号增强的效果。

波束成形怎么知道该如何调整发射天线的信号呢？这是通过检测信道状态信息CSI(Channel State Information)来获取并计算出调整参数的。按照CSI获取方式的不同，波束成形可以分为显式波束成形(Explicit beamforming)和隐式波束成形(Implicit beamforming)。

2.1.2.1 显式波束成形

显式波束成形是一种需要STA反馈信道信息的波束成形方式。显式波束成形信道信息的探测和反馈流程如下：

• AP向STA发送探测数据(Training Symbol)。在802.11n标准中，AP发送探测数据包括空数据包(NDP)和交错前导码两种方式；802.11ac标准开始则直接使用NDP的方式。

• NDP是一种没有数据的空帧，其没有负荷。AP向STA发送NDP探测通告和空数据包，STA收到探测通告和空数据包后进行信道信息反馈。

• 交错前导码是利用发送的带有负荷的帧实现探测，这个帧会承载一个MAC帧和探测信道。

• STA向AP反馈信道信息。

在802.11n标准中，STA的信道信息反馈有三种方式：

• CSI方式：STA将原始的信道信息直接发送给AP，由AP计算最终的波束成形权重值。

• 非压缩波束成形权重值：STA收到探测后，由STA计算出波束成形权重值后反馈给AP。这种方式会增加系统开销，因此有了压缩波束成形权重值。

• 压缩波束成形权重值：同样是由STA计算出波束成形权重值后反馈给AP，并且通过一些方法降低了系统开销。

802.11ac标准开始，STA采用压缩波束成形权重值的方式反馈信道信息。

• AP根据STA反馈或自己计算的权重信息进行波束成形，多径信号在STA处汇聚，形成增益。

2.1.2.2 隐式波束成形

隐式波束成形是由Beamformer计算发送方向信道信息的波束成形方式，其利用了时分双工(TDD，Time Division Duplexing)系统的互易性(即认为同频段的上下行的信道状态信息是相等的)，将Beamformee反馈的上行接收方向信道信息直接应用于下行发送方向，进行波束成形。由于隐式波束成形具体实现的复杂性，从802.11ac协议开始已经明确不再支持隐式波束成形且业界也基本无厂商实现该方式波束成形，这里我们也不再做更详细的描述。

2.2 Single-User (SU) Beamforming

Single-user beamforming is readily understandable because its purpose is to shape a transmission from a single transmitter to a single receiver. As shown in Figure, the beamformer sends a null data packet, which is a frame with a known fixed format. By analyzing the received NDP frame, the beamformee calculates a feedback matrix that is sent in a reply frame. Beamformees do not send a steering matrix directly because the beamforming sounding protocol needs to enable multiple-user MIMO, as described in the next section.

2.2.1 Channel Calibration for Single-User Beamforming

The channel calibration procedure is carried out as a single operation, in which the beamformer and beamformee cooperatively measure the channel to provide the raw data needed to calculate the steering matrix. The sounding procedure does not transmit the steering matrix directly, but instead works to exchange all the information necessary for the beamformer to calculate its own steering matrix.

2.2.1.1 NDP Announcement frame

The channel sounding process begins when the beamformer transmits a Null Data Packet Announcement frame, which is a control frame and is depicted in Figure. The entire channel sounding process is carried out in one burst, so the duration set in an NDP Announcement corresponds to the length of the full exchange of three frames. In single-user MIMO beamforming, the NDP Announcement frame relays the size of the feedback matrix by identifying the number of columns in the feedback matrix.

The main purpose of the NDP Announcement frame is to carry a single STA Info field for the intended beamformee. The STA Info field is two bytes long and consists of three fields:

AID12 (12 least significant bits of the intended beamformee’s association ID)

Upon association to an 802.11 access point, client devices are assigned an association ID. The least significant 12 bits of the beamformee’s association ID are included in this field. When a client device acts as a beamformer, this field is set to 0 because the AP does not have an association ID.

Feedback Type

In a single-user NDP Announcement frame, this field is always 0.

Nc Index

This index describes the number of columns in the feedback matrix, with one column for each spatial stream. As a three-bit field it can take on eight values, which matches the eight streams supported by 802.11ac. This field is set to the number of spatial streams minus one.

2.2.1.2 NDP frame

Upon transmission of the NDP Announcement frame, the beamformer next transmits a Null Data Packet frame, which is shown in Figure. The reason for the name “null data packet” should be obvious in looking at the frame; Figure shows a PLCP frame with no data field, so there is no 802.11 MAC frame. Channel sounding can be carried out by analyzing the received training symbols in the PLCP header, so no MAC data is required in an NDP. Within an NDP there is one VHT Long Training Field (VHT-LTF) for each spatial stream used in transmission, and hence in the beamformed data transmission.

2.2.1.3 VHT Compressed Beamforming Action frame

Following receipt of the NDP, the beamformee responds with a feedback matrix. The feedback matrix tells the beamformer how the training symbols in the NDP were received, and therefore how the beamformer should steer the frame to the beamformee. Figure shows the format of the compressed beamforming report frame used in single-user MIMO. The Action frame header indicates that the frame contains a feedback matrix.

The VHT MIMO Control field, included next, enables a beamformer to interpret the feedback matrix by describing the following attributes:

Size of the feedback matrix (6 bits)

The Nc Index and Nr Index fields describe the size of the feedback matrix in terms of the number of columns and the number of rows. When using beamformed transmissions over large numbers of spatial streams, the matrix will be quite large.

Channel width (2 bits)

The feedback matrix’s size also depends on the size of the underlying channel. Wider channels require larger feedback matrices because there are more individual carriers to measure.

Grouping (2 bits)

When parts of the beamforming matrix are repeated, the beamformee can group multiple spatial streams together to reduce the size of the transmitted matrix.

Codebook (1 bit)

Roughly speaking, a beamforming matrix is used to describe the phase shifts required by each antenna element (see Figure). 802.11ac transmits the information on these angles as a long string of bits; the receiver of a steering matrix needs to know where to split the bit field into individual matrix elements, and this field is used to describe the representation of the data.

Type of feedback (1 bit)

Obviously, in single-user MIMO, the feedback type will be single user.

Flow control (10 bits)

The Remaining Feedback Segments and First Feedback Segment fields are used together with the Sounding Dialog Token to match the response from the beamformee to the beamformer’s request. In very large matrices with wide bandwidths and high numbers of spatial streams, the matrix will be quite large and therefore may need to be sent to the beamformer in multiple steps.

2.3 Multi-User (MU) Beamforming

By simplifying beamforming to use one method of channel sounding, 802.11ac will enable wider use of standards-based beamforming. More significant, however, is the inclusion of multi-user (MU) MIMO beamforming in 802.11ac. Prior to the introduction of multi-user beamforming, all 802.11 devices could send a transmission to only one device at a time. Just as Ethernet switches reduced the scope for collisions from a large network down to a single port, multi-user MIMO reduces the spatial collision domain. By using MU-MIMO, an AP may transmit to multiple receiving stations simultaneously.

One important capability that MU-MIMO brings to 802.11ac is its support of single-stream devices. Prior to 802.11ac, beamforming worked to increase the signal-to-noise ratio of a link to a single device, but the devices on the network often limited its benefits. Many small battery-powered devices are capable of only a single spatial stream, and thus receive only limited benefits from single-user MIMO. With 802.11ac’s multi-user MIMO, a single transmission time can be used to send frames to multiple single-stream receivers. The 802.11ac standard allows up to four different receiver groups within one MU-MIMO transmission.

2.3.1 Channel Calibration for Multi-User Beamforming

To support multi-user MIMO beamforming, 802.11ac uses an extended version of the channel sounding exchange. As shown in Figure, the multi-user channel sounding procedure requires a response from all beamformees. Each beamformee contributes information in a feedback matrix, and the beamformer uses multiple feedback matrices to produce one steering matrix.

In Figure, the sounding procedure starts off exactly as it did in the single-user case, with an NDP Announcement and NDP that put the transmission out to begin the calibration. However, to retrieve the feedback matrix from each beamformee, the multi-user sounding procedure needs a new frame, the Beamforming Report Poll frame, to ensure that responses from all beamformees are collected. Figure  shows three beamformees, and therefore the beamformer must use two poll frames to obtain the feedback matrices from the second and third beamformees. (No poll frame is required for the first station named in the NDP Announcement frame, but the second and subsequent beamformees must be polled.) After receiving multiple responses, the beamformer will integrate all the responses together into a master steering matrix.

2.3.1.1 NDP Announcement frame

Upon transmission of the NDP Announcement frame, the beamformer next transmits a Null Data Packet frame. Like NDP Announcements, null data packets are sent in single-user mode. Therefore, even in multi-user MIMO sounding, the format will be identical to the single-user format shown in Figure. A single null data packet has no MAC header information and will be received by all devices. Each device can use the received training frames in the null data packet to calculate its feedback matrix.

2.3.1.2 Compressed Beamforming Action frame

The Compressed Beamforming Action frame serves the same purpose in multi-user MIMO as it does in single-user MIMO. However, the multi-user format of the frame includes an extra field, the Multi-User Exclusive Beamforming Report field, at the end of the frame. This field carries signal-to-noise ratio differences between subcarriers and is needed to update the steering matrix when there are multiple recipients. Both report fields shown in Figure are indicated as variable because their size depends on the number of spatial streams as well as the channel bandwidth.

2.3.1.3 Beamforming Report Poll frame

To retrieve additional feedback from the second and subsequent beamformees, the beamformer must use the Beamforming Report Poll frame, which is a control frame. This frame is quite simple, as can be seen in Figure: it is essentially a one-byte field of retransmission requests. Each bit in the Feedback Bitmap field requests one feedback segment to be retransmitted.

3. Abbreviations and acronyms

• BF(BeamForming, 波束成形)
  • SU Beamformer(发射端的单用户波束成形)
  • SU Beamformee(接收端的单用户波束成形)
  • MU Beamformer(发射端多用户波束成形)
  • MU Beamformee(接收端的多户波束成形)
• TxBF(Transmit beamforming, 发射波束成形)
• CSI(Channel State Information)
• TRQ(training request)
• NDP(null data PPDU)
• HTC(high-throughput control)
• PPDU(PLCP Protocol Data Unit, PLCP协议数据单元)