802.11ba (WUR) or how to cross a snake with a hedgehog

Not so long ago, on all sorts of other resources and on my blog, I talked about the fact that ZigBee is dead and it’s time to bury the stewardess. In order to make a good face in a bad game with Thread working on top of IPv6 and 6LowPan, which is more suited to this Bluetooth (LE). But I’ll talk about this some other time. Today we will talk about how the committee worked outthey thought well for the second time after 802.11ah and decided that it was time to add a full-fledged version of something like LRLP (Long-Range Low-Power) to the 802.11 standards pool similar to LoRA. But this was not feasible without killing the sacred backward compatibility cow. As a result, Long-Range was abandoned and only Low-Power remained, which is also very good. The result was a mixture of 802.11 + 802.15.4 or simply Wi-Fi + ZigBee. That is, we can say that the new technology is not a competitor to LoraWAN solutions, but rather is created in order to complement them.

So, let's start with the most important thing - Now in devices with support for 802.11ba there should be two radio modules. Apparently, looking at 802.11ah / ax with their Target Wake Time (TWT) technology, the engineers decided that this was not enough and that energy consumption should be drastically reduced. Why the standard provides for separation into two different types of radio - Primary Communication Radio (PCR) and Wake-Up Radio (WUR). If everything is clear with the first, here it is the main radio, it transmits and receives data, but with the second it is not very. In fact, WUR is for the most part a listening device (RX) and, by design, should consume very little power to operate. Its main task is to receive an awakening signal from the AP and turn on the PCR. I.e,this method significantly reduces the cold start time and allows you to wake up the device at a given time with maximum accuracy. This is very useful when you have, say, not ten devices, but one hundred and ten and you need to exchange data with each of them in a short period. Plus, the logic of the frequency and periodicity of awakening moves to the AP side. If, say, the PUSH methodology is applied in LoRAWAN when the actuators themselves wake up and broadcast something and sleep the rest of the time, then in this case, on the contrary, the AP decides when and which device should wake up, and the actuators themselves ... not always are sleeping.the logic of the frequency and frequency of waking moves to the AP side. If, say, the PUSH methodology is applied in LoRAWAN when the actuators themselves wake up and broadcast something and sleep the rest of the time, then in this case, on the contrary, the AP decides when and which device should wake up, and the actuators themselves ... not always are sleeping.the logic of the frequency and frequency of waking moves to the AP side. If, say, the PUSH methodology is applied in LoRAWAN when the actuators themselves wake up and broadcast something and sleep the rest of the time, then in this case, on the contrary, the AP decides when and which device should wake up, and the actuators themselves ... not always are sleeping.

Now let's move on to frame formats and ensuring compatibility. If 802.11ah as the first attempt was created for the 868/915 MHz bands or simply SUB-1GHz, then 802.11ba is already intended for the 2.4GHz and 5GHz bands. In previous “new” standards, compatibility was achieved through a preamble that was understandable to older devices. That is, the calculation has always been that old devices do not have to be able to recognize the entire frame, they just need to understand when this frame will begin and how long the transmission will take. It is this information that they take from the preamble. 802.11ba was no exception, since the circuit is proven and tested (we will omit the cost issue for now).

As a result, the 802.11ba frame looks something like this:



The non-HT preamble and a short fragment of OFDM with BPSK modulation allows all 802.11a / g / n / ac / ax devices to hear the beginning of the transmission of this frame and not to intervene, going into listening mode. The preamble is followed by a synchronization field (SYNC), which is essentially an analog of L-STF / L-LTF. It serves to give the opportunity to adjust the frequency and synchronize the receiver of the device. And just at that moment, the transmitting device switches to a different channel width of 4MHz. What for? Everything is very simple. This is necessary in order to reduce power and achieve a comparable signal-to-power-to-noise ratio (SINR). Or leave the power as is and achieve a significant increase in transmission distance. I would say that this is a very elegant solution, in addition, which can significantly reduce the requirements for power supplies. Recalle.g. popular ESP8266. In transmission mode using a bitrate of 54 Mbps and a power of 16dBm, it consumes 196 mA, which is an overwhelming amount for something like CR2032. If we reduce the channel width by a factor of five and reduce the transmitter power by a factor of five, then we will practically not lose in transmission distance, but it will be possible to reduce the current consumption by several times, say, to about 50 mA. It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.popular ESP8266. In transmission mode using a bitrate of 54 Mbps and a power of 16dBm, it consumes 196 mA, which is an overwhelming amount for something like CR2032. If we reduce the channel width by a factor of five and reduce the transmitter power by a factor of five, then we will practically not lose in transmission distance, but it will be possible to reduce the current consumption by several times, say, to about 50 mA. It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.popular ESP8266. In transmission mode using a bitrate of 54 Mbps and a power of 16dBm, it consumes 196 mA, which is an overwhelming amount for something like CR2032. If we reduce the channel width by a factor of five and reduce the transmitter power by a factor of five, then we will practically not lose in transmission distance, but it will be possible to reduce the current consumption by several times, say, to about 50 mA. It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.In transmission mode using a bitrate of 54 Mbps and a power of 16dBm, it consumes 196 mA, which is an overwhelming amount for something like CR2032. If we reduce the channel width by a factor of five and reduce the transmitter power by a factor of five, then we will practically not lose in transmission distance, but it will be possible to reduce the current consumption by several times, say, to about 50 mA. It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.In transmission mode using a bitrate of 54 Mbps and a power of 16dBm, it consumes 196 mA, which is an overwhelming amount for something like CR2032. If we reduce the channel width by a factor of five and reduce the transmitter power by a factor of five, then we will practically not lose in transmission distance, but it will be possible to reduce the current consumption by several times, say, to about 50 mA. It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.which is prohibitively large for something like CR2032. If we reduce the channel width by a factor of five and reduce the transmitter power by a factor of five, then we will practically not lose in transmission distance, but it will be possible to reduce the current consumption by several times, say, to about 50 mA. It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.which is prohibitively large for something like CR2032. If we reduce the channel width by a factor of five and reduce the transmitter power by a factor of five, then we will practically not lose in transmission distance, but it will be possible to reduce the current consumption by several times, say, to about 50 mA. It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.If we reduce the channel width by a factor of five and reduce the transmitter power by a factor of five, then we will practically not lose in transmission distance, but it will be possible to reduce the current consumption by several times, say, to about 50 mA. It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.If we reduce the channel width by a factor of five and reduce the transmitter power by a factor of five, then we will practically not lose in transmission distance, but it will be possible to reduce the current consumption by several times, say, to about 50 mA. It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.It’s not that it was critical from the side of the AP that sends the frame to WUR, but still not bad. But for STA this already makes sense, since lower consumption allows you to use just something like CR2032 or batteries that are used for long-term energy storage with low rated discharge currents. Of course, nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.nothing happens for free and a decrease in channel width will lead to a decrease in channel speed with an increase in the transmission time of one frame, respectively.

Speaking of channel speed. The standard in its current form provides two options: 62.5 Kbps and 250 Kbps. Feel ZigBee smelled? This is not casual, since it has a channel width of 2Mhz instead of 4Mhz, but a different type of modulation with a higher spectral density. As a result, the radius of action of 802.11ba devices should be longer, which is very useful for indoor IoT scenarios.

Although, wait a minute ... Keep all stations in the district silent, using only 4 MHz from the 20 MHz band ... "THIS IS THE SAME WASTE!" - you will say and you will be right. But no, THIS IS REAL WASTE!



The standard includes the ability to use 40 MHz and 80 MHz subchannels. At the same time, the bitrates of each subchannel can be different, and in order to match the broadcast time, Padding is added to the end of the frame. That is, the device can take airtime at all 80 MHz, and use it only at 16 MHz. This is a real waste.

By the way, the devices surrounding Wi-Fi have no chance of understanding what is being broadcast there. Because for encoding 802.11ba frames, the OFDM familiar to them is NOT used. Yes, this is how the famously alliance abandoned the fact that it worked without fail for many years. Instead of the classic OFDM, Multi-Carrier (MC) -OOK modulation is used. The 4MHz channel is divided into 16 (?) Subcarriers, each of which uses Manchester coding. At the same time, the DATA field itself is also logically divided into segments of 4 μs or 2 μs depending on the bit rate, and in each such segment, a low or high level of encoding can correspond to a unit. Such a solution to avoid a long sequence of zeros or ones. Minimal scrambling.



MAC level is also extremely simplified. It contains only the following fields:

• Frame Control
  
  Beacon, WuP, Discovery .
  Beacon , WuP , Discovery STA AP , 802.11ba. 48 .
• ID
  
  AP, STA STA . (, , groupcast wake-ups ).
  
• Type Dependent (TD)
  
  A fairly flexible field. It is in it that the exact time can be transmitted, a signal about updating the firmware / configuration with the version number or something useful that the STA should know about.
• Frame Checksum Field (FCS)
  It's simple. This is a checksum.

But in order for the technology to work, it’s not enough just to send a frame of the desired format. STA and AP must agree. STA reports its parameters, including the time it takes to initialize the PCR. All negotiation takes place using regular 802.11 frames, after which the STA can turn off the PCR and go into WUR activation mode. And maybe even a little sleep, if there is such an opportunity. Because if it is, it is better to use it.
Then begins a small squeeze of the precious milliamperes of watches under the name WUR Duty Cycle. There is nothing complicated, just STA and AP, by analogy with there, as it was for TWT, agree on a sleep schedule. After that, the STA mostly sleeps, occasionally including WUR, in order to listen to “Has anything useful come to me?” And only if necessary awakens the main radio module for traffic exchange.

Radically changes the situation compared to TWT and U-APSD, is not it?

And now an important nuance that you do not immediately think about. WUR does not have to run at the same frequency as the main module. On the contrary, it is desirable and recommended that he work on another channel. In this case, the functionality of 802.11ba does not interfere with the operation of the network in any way and vice versa it can be used to send useful information. Location, Neighbor List and much more under other 802.11 standards, such as 802.11k / v. And what advantages open up for Mesh networks ... But this is the topic of a separate article.

As for the fate of the standard itself as a document, at the moment Draft 6.0 is already ready with an Approval rate of 96% . That is, this year we can expect the most real standard, or at least the first implementations. And how much it will gain distribution - only time will tell.

Such things ... (c) EvilWirelesMan .

Recommended reading:

IEEE 802.11ba - Extremely Low Power Wi-Fi for Massive Internet of Things — Challenges, Open Issues, Performance Evaluation

IEEE 802.11ba: Low-Power Wake-Up Radio for Green IoT

IEEE 802.11-Enabled Wake-Up Radio : Use Cases and Applications