802.11be (Wi-Fi 7) Technology Outlook

1. Overview of Wi-Fi 7 New Features

Figure 1 is a schematic diagram of the technical evolution of Wi-Fi technology from Wi-Fi 6 to Wi-Fi 7. In order to achieve the key goals of EHT, the protocol has added many new features.

Figure 1 Technology evolution from Wi-Fi 6 to Wi-Fi 7

The main features of the PHY layer are as follows:

• The maximum bandwidth has been increased to 320MHz. A cleaner 6GHz frequency band has been introduced in Wi-Fi 6E, which will be continued in Wi-Fi 7. Currently, my country plans to use the high-frequency band of 700MHz (6425~7125MHz) for 5G or 6G, and the low-frequency band is still under discussion.
• The modulation method uses 4096-QAM, and each symbol carries 12 bits of information. As the modulation rate increases, the modulation rate MCS12-13 is increased. At the same time, the use of 4K modulation makes beam forming essential, and the transmission EVM (Error Vector Magnitude) must reach -38dB to balance the transmission power and signal distortion.
• The maximum spatial stream is increased to 16x16. In fact, for the AP, this many spatial streams are for MU-MIMO rather than pure MIMO, because the terminal will not add so many antennas considering practicality and cost.
• OFDMA improvements include MRU (Multi Resource Unit) allocation and preamble puncture.
• Others include the newly added fields U-SIG and EHT-SIG in the preamble.

The main characteristics of the MAC layer are:

➢ Multi-AP collaboration involves mutual perception between APs, including collaborative OFDMA, collaborative TDMA, collaborative spatial stream multiplexing, collaborative beamforming, collaborative complementary transmission and other technologies. The current protocol does not specify a detailed solution, so this article will not discuss it.

➢ MLO (Multi-Link Operation), this feature requires the support of MLD (Multi-Link Device) hardware and is the focus of this article.

2. OFDMA Improvement

Orthogonal Frequency Division Multiple Access (OFDMA) is a new technology introduced by Wi-Fi 6, which can effectively improve the utilization of spectrum resources to improve WLAN performance. In previous Wi-Fi technologies, users used channels through different time slices, and in each time slice, users completely occupied all subcarriers to perform a complete transmission and reception. In order to multiplex channels, the 802.11ax standard divides channel resources in the same way as LTE, and calls the smallest subchannel a resource unit RU (Resource Unit). Each RU includes at least 26 subcarriers, and user usage is allocated according to RU resources. In this mode, user data is carried on each RU. From the perspective of the overall time and space domain, multiple users are allowed to use different RUs to transmit and receive data in each time slice, as shown in Figure 2, where different colors represent different users.

Figure 2 OFDM VS. OFDMA

In Wi-Fi 6, each terminal can be assigned only one RU. Although this allocation method simplifies the allocation mechanism, it brings the following usage limitations:

• When the number of users is relatively small, the utilization rate of spectrum resources is limited.
• When preamble puncturing occurs, such as radar signal detection, because channel bundling is used, only the 20MHz bandwidth of the main channel can be used, and other bandwidth resources are wasted.
• Each user can only be assigned to a specific RU, which greatly limits the scheduling of spectrum resources.

Wi-Fi 7 introduces a multi-RU allocation mechanism on this basis, which makes the RU resource allocation for a single terminal more flexible and can be allocated to the terminal in combination under different bandwidths. As shown in Figure 3, under 160MHz bandwidth, for channel puncture under different 40MHz channels, the MRU allocation mechanism can reallocate 996+484 subcarrier RU resource combinations (some subcarriers are used for DC protection). Figure 4 shows the comparison of RU allocation methods of 802.11ax and 802.11be.

Figure 3 160MHz ETH PPDU sending MRU combination method

Figure 4 Comparison of RU allocation methods between 802.11ax and 802.11be

In the above text, we mentioned the concept of preamble puncture. This technology was introduced in Wi-Fi 6. It is based on OFDMA transmission mode and can effectively optimize channel bundling access. In Wi-Fi 7, due to the expansion of spectrum, this technology is used to assist MRU allocation. We use Figure 5 to simply illustrate the technical details. In Figure 5, four channels 52, 56, 60, and 64 are bundled into an 80MHz channel, with channel 52 as the main channel and the others as secondary channels. In Wi-Fi 5 (802.11ac), if a radar signal is detected in channel 56, the terminal can only transmit in the main channel 52. In Wi-Fi 6, the AP and the terminal can close the frequency of channel 56 and communicate using the remaining three channels. Even if it still works in the 80MHz channel mode, the actual transmission sets channel 52 to NULL, and does not interfere with other radios in channel 52. It should be noted that 802.11ax stipulates that the main channel cannot be shielded. The 802.11ax protocol has made specifications for shielded channels, and 802.11be has made changes to this specification.

Figure 5 Schematic diagram of channel puncture

3. MLO Mechanism

In 802.11be, a major change in PHY is MLD, which is the hardware that provides multi-link physical layer support. The MAC part of MLD is MLO. Although ICs (simply understood as radios) before MLD support connections in multiple frequency bands, they can only select one frequency band for connection at a time. For a terminal, only a single Wi-Fi connection can be established with the AP at a time, that is, it is either connected to 2.4GHz or 5GHz. Multiple APs can work on multiple frequency bands at the same time, but the implementation method is to process them through multiple ICs, that is, different frequency bands are isolated using different ICs, so that multiple frequency bands can work in parallel without affecting each other.

The reason why an IC can only provide a connection in one frequency band is largely related to the setting that only a single baseband is provided inside the IC. This design is mainly to save costs and reduce design difficulty. In the past, terminal Wi-Fi mainly focused on stable network connection, while throughput, latency and jitter were not so critical. However, with the rapid development of the network, the demand for throughput and latency has increased, so the multi-link working mode has also emerged.

Figure 6 AP and STA multi-link connection

In the figure above, MLD refers to a device that provides multiple links. It can be understood that AP1, AP2, and AP3 are one or more RF ICs (Radio) in the same AP. This RF IC contains three frequency bands, corresponding to 2.4GHz, 5GHz, and 6GHz. The lower half of the figure corresponds to the terminal STA. Like the AP, the STA also has a RF IC corresponding to three frequency bands. Multiple connections can be established between the AP and the STA, namely links 1, 2, and 3 in the figure. This feature of Wi-Fi 7 can optimize the use of spectrum resources through multiple links, bringing the following advantages:

• Increased throughput: Multiple links mean that data can be transmitted simultaneously. Regardless of the transmission mode, the throughput is equivalent to the sum of all links.
• Reduced latency: Due to the existence of multiple links in different frequency bands, compared with the previous single-link channel competition, multi-link increases the probability of devices obtaining channel communication, thereby reducing communication latency. At the same time, the selectivity of terminal access is also greatly improved.
• Increase data transmission reliability: Different links can transmit the same data, and the links can complement each other, reducing the retransmission of data frames and improving data transmission reliability.
• Transmission splitting and isolation: Different links operate on different frequency bands, and when multiple links are used, they support simultaneous bidirectional transmission, that is, link 1 sends and link 2 receives at the same time.

3.1 MLO link connection mode

According to the link connection method, MLO can be divided into the following five categories, as shown in Table 1:

Table 1 MLO link classification

Figure 7 shows the MLO mode of single-radio multi-link. In the MLSR (Multi-Link Single Radio) working mode, the STA has only one radio but can switch different channels for connection. Only one of the two links can be used for TX (transmit)/RX (Receive) at the same time.

Figure 7 Single radio multiple links

Figure 8 shows the single-radio multi-link enhanced MLO mode. In the EMLSR (Enhanced MLSR) working mode, the STA has only one radio, but its 2x2 spatial stream module can be configured to work separately in 1x1 working mode, and can track and monitor two channels at the same time. According to the channel congestion situation, it switches to the corresponding channel link for TX/RX.

Figure 8 Single radio multi-link enhancement

Figure 9 shows the multi-radio multi-link MLO mode. In the EMLMR (Enhanced Multi-Link Multi-Radio) working mode, the 2x2 spatial streams of the two radios of the STA can be configured and combined. The Tx/Rx of Radio1 works on link 1, and the Tx/Rx of Radio2 works on link 2. It also supports the combination of Radio1 Rx and Radio2 Rx working on link 1 and Radio1 Tx and Radio2 Tx working on link 2. Finally, data transmission is performed on a certain link according to the congestion of the channel where the link is located.

Figure 9 Multi-radio multi-link enhancement

Figures 10 and 11 show the asynchronous and synchronous modes of multi-radio multi-link. For the multi-link of MLMR, because there are multiple radios, the use of the corresponding links is divided into synchronous and asynchronous modes, namely NSTR MLMR and STR MLMR. Although different links have different operating frequency bands, the spacing of RF modules and other issues will cause inter-module interference (IDC: In-Device Coexistebce Interference). The two types of methods will also lead to different channel access and usage mechanisms. Currently, there is little introduction to the two methods in the protocol, so this article will briefly introduce them.

The full name of STR asynchronous mode is Simultaneous Tx and Rx, which is a simultaneous transmit and receive working mode, but it is not a full-duplex mode. It means that different links are allowed to work at the same time, and the same channel still needs to compete for the sending window. Different links work on different channels, do not interfere with each other, and work independently, so when in use, there will be a state where link 1 sends and link 2 receives, so it is called asynchronous mode. STR actually means that different radios of the terminal run on different frequency bands, but the following issues still need to be considered in implementation:

IDC (In-Device Coexistence Interference) issues, the channel isolation between different frequency bands needs to be considered. The subsequent 802.11be group discussion may develop a detailed channel isolation division, which will be different when 2.4GHz and 6GHz are used at the same time, 5GHz and 6GHz are used at the same time, or 2.4+5+6GHz are used.

The use scenarios of multi-links in asynchronous situations are expanded. For example, the retransmission method can be used to increase transmission reliability, allowing link 1 to retransmit to link 2. This can effectively avoid interference with a channel, or use a link for tasks such as control channels.

Figure 10 Asynchronous multi-radio multi-link, allowing multiple links to work in both directions simultaneously

Figure 11 Synchronous multi-radio links, does not support independent operation of multiple links

The NSTR synchronization mode requires that the transmission process starts and ends at the same time. This rule is mainly based on the fact that the ACK feedback mechanism is direct feedback rather than interactive feedback (BAR and BA are typical interactive feedback). The ACK feedback method will cause two problems. First, if the transmission does not end at the same time, one link will have fed back ACK while the other link is still transmitting. Second, the ACK timeout mechanism. If the two links do not end at the same time, the timeout time of each link will be inconsistent. For the problem of simultaneous transmission, the current reference is mainly several discussion documents of the 802.11be group. In "11-20-0993-04-00be-sync-ml-operations-of-non-str-device", several implementation methods of synchronous MLO are mentioned, as shown in Figures 12, 13, and 14. I am currently in a state of superficial understanding, so I can only give a simple explanation of the discussion content.

Figure 12 NSTR Scheme 1 - PIFS

Solution 1 adopts the PIFS mechanism in channel competition. Both links compete for channels. The link that wins the channel first triggers the transmission process of the other link. The premise is that the link that fails to compete for the channel needs to perform PIFS to check whether it is idle, omitting the back-off process in the channel competition of the link.

Solution 2 is the ePIFS (enhanced PIFS) mechanism, which is an enhancement of Solution 1. It adds a certain backoff time slot for the link that has not competed for the channel, but does not set the NAV value of the link channel. It waits for the priority competing link to synchronously trigger the direct use of the channel.

Figure 13 NSTR Scheme 2 - ePIFS

In solution 3, a link waiting time slot is added, which is relatively fair to both non-Wi-Fi 7 terminals and Wi-Fi 7 terminals. Both links compete for the channel normally, but the time of simultaneous transmission is based on the link that competes later. The link that competes first increases the backoff time to wait for simultaneous transmission.

Figure 14 NTSR Solution 3 - Wait Slot

Currently, options 1, 2, and 3 all have various unexpected problems. According to the conclusion of the group discussion, the probability of selecting option 1 is higher.

3.2 MLO Service Discovery

Wireless service discovery includes active scanning and passive scanning. The change in the service discovery method under the MLO mechanism is that STA is required to scan in different link channels, which will cause the scanning time to increase exponentially, so some mechanisms need to be introduced to simplify the process.

Each AP that supports MLD needs to carry the basic connection information RNR (Reduced Neighbor Report) of other radios in the Beacon or Probe Rsp frames sent by each of its radios, including country code, BSSID, SSID, BSS parameters, etc. At the same time, the STA will send an ML Probe Req frame to obtain the complete connection information of all AP radios. When the AP receives it, it will reply with the corresponding ML Probe Rsp frame, which contains the multi-link information set in the AP. The overall simplified process is shown in Figure 15.

Figure 15 MLO service discovery process

3.3 MLO association and authentication

After STA service discovery, the next step is association and authentication. The previous association method only supports the association process of one link, so three links require three association processes. The current reassociation request and reassociation response need to add new fields based on the original frame structure, mainly including: Common Info Field and information of each Radio (Per-STA profile), as shown in Figure 16.

Figure 16 MLO association method

The authentication method of MLO is shown in Figure 17, which is basically the same as the normal four-way handshake key exchange, with additional new keys. PMK and PTK are obtained normally; PMK, PTK and PN space are filled by PTKSA, and different links use different GTK, IGTK and BIGTK.

Figure 17 MLO authentication method

3.4 Other changes in MLO features

Other key features of MLO include:

• BA is used in each link.
• The reception status of the QoS data frame can be fed back through other links.
• The TIM domain in Beacon contains the message cache status of the entire link, as shown in Figure 18.
• In the default mode, each link contains all TID mappings. In the optional mode, each link can be set with different TID mappings, but there is at least one TID mapping when a link state is available, as shown in Figure 19.

Figure 18 TIM domain collection in Beacon

Figure 19 TID mapping of each MLO link

IV. Conclusion

Due to the limited space and my understanding, this article only briefly introduces some of the contents in the 802.11be draft D1.5. The current 802.11be protocol standard is still under revision, and some of the contents in this article may be updated or revised. I personally believe that Wi-Fi 7 under the new Wi-Fi standard will allow consumers to experience a huge leap in wireless performance. From 8K streaming video to immersive VR experience, users around the world will enjoy unprecedented ultra-high speed and ultra-low latency experience. Let us wait and see.