Hey.
If you believe Einstein's theory of simplicity, the main indicator of understanding a subject is the ability to explain it as simply as possible, then in this post I will try to explain the effect of just one detail of the new standard as simply and in detail as possible, which for some reason even the Wi-Fi Alliance considers unworthy of mention in infographic about the new features of Wi-Fi 6, although, as we will soon see together, it is very important and remarkable. Not everything here is deep enough and certainly not comprehensive (because such an elephant is difficult to eat even in parts), but I hope that we will all learn something new and interesting for ourselves from my verbal exercises.
The same 802.11ax, which we have been waiting for from day to day for at least the second year, carries a lot of new and amazing things. Everyone who wants to tell something about him is always faced with a choice: either arrange overview jumps over the heads, mentioning a bucket of abbreviations and abbreviations, trying not to get bogged down in the complex mechanisms under the hood of each of them, or wrap up an hour-long report about one thing most pleasing to the author. I will venture to go even further: most of my note will be devoted to not even new things!
So, for more than twenty years, some of the wireless data networks have been built according to a bunch of standards of the 802.11 family, and, like any self-respecting speaker, I would have to slightly restore the timeline of the entire event chain that gave the world billions of devices compatible with each other - but , as an author respecting the reader, I will risk not doing this. However, something should be reminded to each other.
All iterations of Wi-Fi have prioritized reliability rather than maximizing throughput. This follows from the media access mechanism (CSMA / CA), which is not the most optimal in terms of squeezing the last kilobits per second from the transmission medium (more details about the imperfection of the world in general and Wi-Fi in particular, if you wish, can be found in the article of my former colleague ), but incredibly tenacious in almost any conditions. In fact, you can break almost all the basics of Wi-Fi network design - and in such a network data will still be exchanged! What in English is called robustness is the whole mechanism by which Wi-Fi clients are able to transmit and/or receive their portions of data. The whole layer of modulation boosts, aggregation of data frames (not exactly the same, but let it be!) continues to work after two basic 802.11 principles that provide this unsurpassed reliability:
- “While one speaks, the rest are silent”;
- “Everything but data is spoken slowly and distinctly.”
The second point causes much more damage to network bandwidth than it might seem at first glance. Here is a cool picture illustrating one sent piece of data on a Wi-Fi network:
Let's see what it means for ordinary people who do not know how many pages are in the 802.11-2016 standard. The data transfer rate that the system writes in the properties of the wireless network and that marketers of any manufacturer draw on access point boxes (well, you probably saw it - 1,7 Gb / s! 2,4 Gb / s! 9000 Gb / s!), not only is the peak and maximum at 100% of busy transmission time, but also is the speed at which only the blue part of this beautiful graph will be sent. Everything else will be sent at a speed, which in English is called the management rate (and in Russian too, because translating such expressions threatens further misunderstanding between engineers), and which is lower not just at times, but in HUNDREDS once. For example, without any additional settings, a network on 802.11ac, which can work with clients at a channel speed of 1300 Mb / s, transmits all service information (everything that is not blue on our increasingly scary graph) at a management rate of 6 Mb / s . More than two hundred times slower!
The logical question is, excuse me, on what date of what month could such a wrecking idea even get into the standard by which billions of devices around the world work? The logical answer is compatibility, compatibility, compatibility! The network on the latest access point should provide the ability to work for ten- and even fifteen-year-old devices, and it is in all these “non-blue” pieces that information flies that slow older devices will hear, correctly understand and will not try to transmit during ultra-fast data pieces their. Robustness requires sacrifice!
Now I am ready to give everyone who is interested in an indispensable tool in order to be horrified by the potential transmitted megabits that are aimlessly lost in modern Wi-Fi - this has already become mandatory for study in the involved engineering circles The WiFi AirTime Calculator by Norwegian 802.11 enthusiast Gjermund Raaen. It is available at - the result of his work looks something like this:
Line 1 is the time taken to transmit a 1512-byte data packet by an 802.11n device in a 20 MHz channel width.
Line 2 is the time taken to transmit the same packet by a device with the same antenna formula, but already operating according to the 802.11ac standard in an 80 MHz channel.
How is it - four times more air is “spoiled”, the maximum modulation has become more complicated from 64QAM to 256QAM, the channel speed is higher in SIX times (433 Mb / s instead of 72 Mb / s), but won from the strength of 25% of the busy time of the air?
Compatibility and the two principles of 802.11, remember?
Well, how can such injustice and wastefulness be corrected - we ask ourselves, as every IEEE working group that started to create a standard probably asked itself? Several logical paths come to mind:
- Accelerate data transfer in the “green” piece of the graph. This is done with the release of each standard, because large numbers look beautiful on the boxes. In practice, as we just noticed, it gives a final increase - even if we accelerate the channel speed to one hundred thousand million gigabits per nanosecond, all other parts of the graph will not go anywhere. That is why I recommend that in all the stories about all the new 802.11 standards, skip the paragraphs where megabits per second are mentioned.
- Accelerate all other parts of the graph. Indeed, if we at least double the speed at which everything “non-green” is transmitted (well, or “non-blue”, if you are still looking at the previous picture), then we will get a little less than 50% increase in real throughput - however, by losing compatibility with devices and a number of other nuances that you will learn about when you go to prepare for the exam for the proud title of CWNA 🙂 Spoiler: it will not always work out, after thinking hard and understanding what it will lead to. In fact, this is a violation of one of the two principles of 802.11, so you need to be very careful with it!
- Blind together several such frames with green parts together. The longer the green part, the more efficient the link rate increase works. Yes, it's a fully working strategy that dates back to 802.11n and is one of several cornerstones of its revolutionary nature. The only problem is that, firstly, a number of applications did not care about such aggregation (for example, the very bloodthirsty Voice over Wi-Fi), and secondly, a number of devices also did not care about it (somehow I decided to catch it though there would be several such aggregated frames on the real network of the company I work in, but for > 500k "captured" frames aggregated of them was exactly zero.Most likely, the problem is in my data collection methodology, but I am ready to discuss it with anyone who wants to somewhere - or in a personal conversation!).
- Violate the first of the two principles of 802.11 by speaking when someone else is speaking. And here, in fact, 802.11ax comes to the rescue.
How cool, finally, in my story about Wi-Fi 6, I got to Wi-Fi 6 itself! If you are still reading this, then you either have to do it for some reason, or you are really interested. So, even though 802.11ax inherits a huge part of the previous developments of the entire 802.11 family (and not only, by the way - some cool things appeared in general in 802.16, aka WiMAX), something in it is still fresh and original. Usually, these words are molded with the following picture, available on the Wi-Fi Alliance website:
As I said from the very beginning, we will be able to consider only one of these key points well enough within one readable article, or rather, none of the ones shown in the picture (what a surprise!). I'm sure you've already read a million cursory descriptions of each of these eight key elements, but I will continue my tediously long story about what follows from OFDMA - about multiple media access (MU-access control), which, as we see, didn't make it to the infographic. And absolutely in vain!
Multiple access is something without which the division of the channel into subcarriers is generally meaningless. Why try to look at different pieces of the spectrum if there is no mechanism that can force the clients of the new Wi-Fi 6 network to break one of the rules that have not been unshakable up to this point and start talking at the same time? And, of course, such a mechanism simply had to appear - and reduce the impact of the problem of "long" in comparison with the data of service information. How? Yes, it’s very simple: let the “slow” service part be sent in the same way as before, but the “fast” part, in which the data goes directly, we will send simultaneously from several (or to several) devices on command! It looks something like this:
It looks complicated, but in essence it is quite easy to explain: using a special frame, understandable to all (not even Wi-Fi 6!) devices, the access point reports that it is ready to transmit data simultaneously to STA1 and STA2. Since the “header” of this frame is completely understandable even to very, very old clients, they make the correct conclusion that the air will be busy for a certain time transmitting information to other network clients, and begin to count down the time until the end of this period (in fact, as always in Wi -fi). But the STA1 and STA2 devices understand that now the data will be transmitted to them in a new way, at the same time, to each on its own piece of the channel, and they also respond to the access point simultaneously, and then just as synchronously confirm the reception of the frame (each with its own portion of data!) , and the environment is released again. “Bottom up” it works in much the same way:
The main and conspicuous difference is the access point, and in this situation it tells the stations that can speak at the same time when to start transmitting, using a special frame, which is called Trigger. This is, in fact, a new “trigger” for the entire mechanism of multiple simultaneous access to the environment, which, in my humble opinion, is one of the most important innovations “under the hood” of the new standard. It is in it that clients receive a “schedule” on how to share one frequency channel among themselves; it is in it that clients simultaneously inform the access point that they received their portions of data and were able to parse them. In it, the access point notifies everyone who can “talk” at the same time about the start of data transfer - the access point is in it and starts sending the required data to it. The new Trigger frame mechanism, in fact, allows you to reduce the irrational use of air occupancy - and as efficiently as many clients can use it and perceive it correctly!
And now let's formulate the main theses that follow from this whole long story and claim to be TL;DR:
- Access points of the new 802.11ax standard, even relying on just one of the many innovations, will begin to increase the total throughput of the entire network from second compatible client device! As soon as at least two clients appear who can speak at the same time, then, all other things being equal (I have no reason to assume that drivers for client radio modules will write better than before, which means that the aggregation of “useful” parts of frames, and many other client-specific functions will still not work so well on average in the zoo) they will ALREADY increase the average throughput. So if you are thinking about a new Wi-Fi network, it makes sense to immediately consider the newest and best access points, because even if there are still few clients for them now, the situation will not remain this way for long.
- All the tricks and tricks that are in the arsenal of a good wireless engineer today will remain relevant for a long time - although the mechanism for accessing the environment has been updated, violating the cornerstone principles that have lasted more than 20 years, it still keeps compatibility at the forefront. You still need to cut off “slow” management rates (and you still need to understand why and when), you still need to properly plan the physical layer, because no mechanism at the data link layer will work if there are problems on the physical one. Just an opportunity to do even better.
- Almost all decisions in Wi-Fi 6 are made by the access point. As we can see, it manages client access to the environment by bundling devices together into "periods" of simultaneous operation. Moving a little further to the side, the work of TWT is also completely on the shoulders of the access point. Now the AP must not only “broadcast the network” and store traffic in queues, but also keep a record of all clients, planning how it is more profitable to combine them with each other based on their bandwidth and traffic needs, their batteries and much, much more — I call this process “orchestration”. The algorithms by which the access point will make all these decisions are not regulated, which means that the real quality and structural approach of manufacturers will manifest itself precisely in the development of orchestration algorithms. The more accurately the points predict the needs of clients, the better and more evenly they will be able to combine them into multiple access groups - therefore, the more rationally the air resources will be used and the higher the final throughput of such an access point will be. The algorithm is the last frontier!
- The transition from Wi-Fi 5 to Wi-Fi 6 is, in its essence and importance, as revolutionary as the transition from 802.11g to 802.11n. Then we got multithreading and “payload” aggregation - now we get simultaneous access to the environment and finally working MU-MIMO and Beamforming (firstly, as we know, these are almost the same thing; secondly, the discussion “ why MU-MIMO was invented in 802.11ac, but could not be made to work” - this is the topic of a separate large article 🙂). Both 802.11n and Wi-Fi 6 operate on both bands (2,4 GHz and 5 GHz), unlike their "intermediate" predecessors - truly, "six is the new four"!
A little about the origins of this article
The article was written for a competition held by Huawei (originally published ). In many ways, when writing it, I relied on my own report at the Wireless Conference, which was held in 2019 in St. Petersburg (a recording of the speech can be viewed , just keep in mind - the sound there, frankly, is not a fountain, despite the St. Petersburg origin of the video!).
Source: habr.com