 Es geht aber heute weniger um mich, es geht um den Lieben Hendrik, er ist ein Netzwerker, er ist Feuerwehrmann und er ist eigentlich auch so ein richtiger WLAN dort, wenn man das so sagen kann. Und er betreut 1600 Access Points, übrigens auch die ganzen Access Points hier im Nock. Dafür mal vielleicht noch eine Runde Applaus. Also die ideale Voraussetzung um uns heute zu erklären. Hallo, dear listeners, welcome to this talk on making Wi-Fi fast again. This is Oliver and Ralph from the translation team. If you have feedback to the translators, treat C3 translate or use the hashtag C33C3T or even us at hello at C3lingo.org. The speaker today is a Wi-Fi expert who has more than 1600 Access Points in operation and he is going to talk about all sorts of Wi-Fi magic today. So please let us welcome Hendrik, our speaker. Hi, also hello from me and welcome to GPN in winter. It's very nice to have you all here. I just heard a boo in terms of Wi-Fi with Nock. We had a small problem. We made Wi-Fi even faster before this talk. And so things had a little hiccup for a while, but now everything should be working again. So the structure of the talk today is I start talking a little bit about me I didn't actually expect somebody to introduce me. Some details for the history of the Wi-Fi standard. How did Wi-Fi develop? When did it come? How long does it exist? A little overview over what changed with IEEE 802.11ac which is the full name of the standard. What changed with this new standard? So then we go more into detail. What changed on layer one of the standard, the physical layer? Because that's really what makes it possible for the higher throughput to appear with the standard. Then I will understand what is this MIMO and this multi-user MIMO. That's very interesting because this again gets a small bandwidth. Then I will go into this magic beamforming that maybe some of you heard about already, which you can do with high frequency waves, but also with audio for example. In the end I will relate this to praxis and reality. How useful is the standard in practice? What does it get us in reality in terms of throughput? Then a slide outlook on the future because IEEE didn't stop working. They are already working on the next standard. So we will see what is coming there. So I'm Hendrik, 23-Jährige. I'm studying in Karlsruhe at the Institute for Electrotechnology. I'm helping out there with the network. I'm responsible for 1600 access points and I'm responsible for setting up the controllers and the installations of the access points in the lecture halls so that in halls like this Wi-Fi is working properly. When I have some free time, I'm also doing amateur radio so playing around. So some little history of 802.11. So this started out really early. They start like cables, that's really nice, but now we come with laptops. And now like already putting your laptop somewhere and plugging in a cable, that's really annoying. Putting the cables everywhere in large cables, that doesn't really work. So at some point they started thinking, okay, let's do this without a cable. Ever since then, in our intervals, they release new standards that each time improve Wi-Fi in some sense, like maybe there's higher throughput or more generally speaking, more efficient Wi-Fi in general. That's the timeline of Wi-Fi here. It started in September 1999 with 802.11a and 802.11b. Those were the really old and slow data rates like 11 megabits per second compared to nowadays, that's like really slow. Back then the idea was, let's have something without a cable and get some data through it. And back in 1999, 909.11, Ambit was actually quite significant, because back then 16 megabit DSL at home, nobody had that. It's not like the DSL speeds we're having today. Then 802.11g was released in June 2003 and then more and more standards appeared, which went further and further and was optimizing data throughput and efficiency, like for example with 802.11g, which maybe you know from your WRT 4.050 GL, which could actually do 54 megabit Wi-Fi, which actually is really cool. And then Fritzbox came and said, oh, I can do 300 megabit and that way standard progressed. Also in the 5 GHz Frequency range, which we have today, back then in 802.11a, those were the first time 5 GHz got into the game. The problem with 5 GHz is through the higher frequency the signal is dammed more strongly through humans or walls. So actually using this to cover a larger area is not really working so well, which is why back then 2.4 GHz was preferred and focused. Because that gives you a higher range, which was more important back then compared to higher throughput. Dann wir got 802.11ac as an latest milestone, which was 2013, released after quite some work. So as a summary I have to say that the 802.11-2007, which appeared in March 2007, that's not really a new standard, it's more like a summary of the existing standards and extension previous prior to that. Because that IEEE standard is being submitted and then all the additional things, all these additional letters, a B, G, A, everything, was just extensions to the basis. So in 2007 they decided, let's just sum this up already so that we have a base standard again in one large block, so that it's easier to read. And if you look through the 802.11ac standard, some things are in italic and it's really just a huge patch for the previous standard. And layering all of these standards on top of each other to build something, it's really hard, which is why in 2007 they just summed it up all in one block. So 802.11ac is always being called this gigabit Wi-Fi and everybody is happy. I can still remember at CBIT, AVM made advertisements, wow, that's really cool. However, the standard is only specified for 5 GHz, because people said, well, 2.4 GHz, we only have 4 channels which we can use effectively without having an overlap. So let's do this 5 GHz only, which is enough, and that makes our job much easier. But then there were additional ways of modulating the wave were picked, which are more efficient, so with one setting of the modulation I will talk more about this later, more bits can be transmitted. We have wider channels, because if the channels are wider then we can send twice as much later because we have twice as much frequency range for the same time. There's fewer MCS values which stands for modulation encoding scheme, which is an index saying which kind of modulation redundancy bit security is used. If you just transmit data you can just transmit it, but you should really expect that your transmission is somehow like there's loss on your transmission and to compensate for that loss you take some of your payload data and add an additional bit or another kind of checksum after it to make sure that actually really everything was transmitted. And these MCS indices are a combination of the modulation so that the bits are secured and probably made redundant. What's also got very interesting is that this beamforming was specified more precisely. In principle beamforming already existed with 802.11n, but there were many different beamforming methods and every manufacturer had their own beamforming method implementing whatever they liked most and not all clients supported it and there were problems with clients and when they tried to do beamforming so with this one they went ahead and said well okay, that's exactly what we're doing now and then as already mentioned previously a multi-user MIMO is coming with 802.11ac which will make us like lots of fun and finally they said that we got 802.11n with 802.11n they made a mistake they defined the standard which was huge it and composes compared to the 54 megabit the 11g which was 11gdude 11n just had way too many things that were added new frequencies and many other things and manufacturers just were not able in the short time to properly implement the standard and provide the hardware so what they said is we're going to have two waves so the first draft version of 11ac when it was released they said this is wave 1 so manufacturers can start producing it and we will guarantee that the part which we released won't be changed in ways such that you will have problems with incompatible clients or that only support the latest version for example and with the second wave that was when the standard was really done it was 2013 and now that's actually the final thing it got interesting with respect to data throughput on the physical layer that is which limits us in the most cases modulation modes and channels we decided on more channels because now the access points don't collide with each other on 2.4 we can use 4 channels without collisions so now we're using more channels we're also using wider channels this increases the throughput fantastically so there's a 3 to 3 MIMO which is being advertised on routers as the number of of aerials normally FAC has defined it and they have defined that there can be up to 8 separate independent transmissions on the same frequency so now we have the 8 8 times the data throughput multi-user MIMO allows simultaneous transmission to multiple users we use separate transmitters inside the access point and channel them to different users we have an improved modulation and coding set and this reorganization of how to do the modulation coding gave them more options this diagram shows all channels that are available not everything is permitted they looked at 5 GHz and then they found there's weather radar and then they found that they have to do dynamic frequency selection they have to check is there a radar and if there's a radar then we better not do anything the access point has to retract from the channel and use something else in Germany the restrictions are even further there was great legislation of what was going to be inside the standard and the regulatory agency both in Europe and in America has added further restrictions this shows the maximum possible and they are fighting for the grey stuff which I can't see so if everything is possible with and without DFS that would be the maximum but it doesn't work everywhere the channels can overlap as in previous wireless land standards this gave more options to avoid collisions between access points changes improvements in the physical layer is the MIMO has been in existence since 802.11. it is if you use 3 aerials then you can use you can use them simultaneously the physical separation on the aerials allows for differentiation its strength of signal and the differentiation of strength of signal allows for different channels one could call it a channel anyway it increases the data throughput there is one data stream per aerial and it allows parallel transmission on the same frequency I will bring a table moment that shows what the 8 partial streams give us in the throughput after the reorganization of the modulation and coding sets they said it was too much and they found that not all of them were used equally we can build better hardware so we don't need all of them so they restricted from the large number of coding sets to a lower number and they have a higher data throughput so these modulation types are available now and here are the improvements the improvement is quadruple amplitude modulation is the QAM and the 256 ones are new the code rate shows how many bits are redundancy or error correction at the bottom you can see that half of all bits are used for protecting the data for long range transmission in other cases they say we can omit the error correction and we have a very good physical conditions and we don't need it and we can use it for throughput quadruple amplitude modulation is a digital modulation a form of modulation and it is a combination of phase modulation and amplitude modulation the I and Q stand for the phase and amplitude und genau die I und Q Werte die für eine QAM Modulation notwendig sind so exaclte the I and Q values are what are not necessary for QAM Modulation this combination can be identified clearly in a big grid and therefore represent the associated bits it depends on how you cover your grid with a bit pattern on how they are mapped on one frequency we provide a small carrier signal and then we add lots and lots of modulated carriers the receiver checks this is what I've received unmodulated carrier as a reference and then I compare the phase and amplitude to the unmodulated one and that gives me the number in higher frequencies we need extra carrier because there's a phase shift in higher frequencies because the frequency is higher and the wave has progressed further already the 64 in the QAM is short for the number of modulation points so this is how I look for example the grid of a QAM64 could look like I is for the in-phase component which is the phase difference Q is the quadrature component 90° angle to the phase 64 values we can encode 6 bit per configuration point so if we for example use the grid code which we can just overlay over this or other encoding methods which you may want to use so for example the 256 QAM uses, which is used with 8.2.11ac uses 2 x 4 bit grid code which means that we got 8 bit that are concatenated and the first 4 bit are gray code which is using the x directions and the remaining 4 bits of the code word are gray code which is oriented in y direction the discussion for how to add gray code on these grids can be extended arbitrarily I recently had a very nice discussion with a roommate over breakfast if you can in an n dimensional space we have m constellation point for every one of these n dimensions and you can have a gray code for every dimension how x is the code word and how many bits y are added with the n to n plus 1 dimension and she threw all sorts of mathematics on top of this, it's possible to do this in the n dimensional but that doesn't really matter for us because we first have to find a third parameter for the space to even use it so I am very happy with the normal qrm software this is a small example let's look at the short point and the top right corner so I just counted from the beginning binary I didn't add any gray code so if you want to have this point that's the 15th point decimal so this is our binary value 4 and the y value of 3 so if my antenna now recognizes I got a phase shift of corresponding to 4 in the x direction and an amplitude difference which corresponds to 3 in the y direction then that's exactly this binary value so that's how this is encoded so this is now we have a huge table over this it starts in the beginning with 802.11n with one spatial stream compared to 802.11ac spatial stream so this table really nicely shows how adding more spatial streams more possibilities for sending more channels on the same frequency so to speak how this increases the data throughput up onto 683 megabit that's way more than the n standard so in its maximal configuration and you still have to as to add in defense of 802.11ac that these gray values these grayways are not even using MCS index 9 they are using MCS 8 because the 20 megahertz channels they must not be used with the MCS index 9 that has been specified like this in the standard which means if you would look at it theoretically which is not allowed by the standard you could pump even more data through this if we now widen the channel then we got even more throughput so then everything is possible again and if we widen the channel again then we get even more throughput and that's the point where things start to get that have holes because 80 megahertz actually not something which was possible in 802.11n however we can widen again we got 160 megabit channels so we get in the bottom right corner we get the value which is the maximally possible throughput rate on 802.11ac that 609 gigabits per second and that we have to start thinking how do I get my data to the access point even with some new standard where I can get 2 or 5 gigabit over my copper wire even then that's not enough so so that was really shooting very high so specify 609 gigabit well let them do that, that's okay and again we have a blue value here the MCS9 is not specified is forbidden for values with 3 streams and 60 megabits for reasons which I don't want to go into here because that's details in the standard and it's not interesting so now this multi-user memo we already have this idea of using the antenna to send multiple streams to the same data which if you think about this, let's already imagine this, we have the same frequency we are sending multiple streams to the same client and they are like somehow separated again, that's already technical genius and then they were like wait, we can do that, we can actually do better, we got the memo with 802.11m but we can go further we don't just have their antenna gain logarithmic in the number of antennas for people who are interested in these details but these people exist already mentioned by room out so they went further and they said we will do concurrent data transmission to all users that we see and we do this just because we want and because we can and then they did so then they had limited slightly so they said let's do no more than four users and let's not do more than four special streams per user but there are no more than eight special streams anyway so this gives us some more advantages for example if we have a laptop which is pulling lots of data right now then if it would be really close to the access point it would be blocking the channel because it would be pulling all the data and then other clients at some point they would also get their turn so there are eight special streams and the laptop is using four special streams so they can do that because the rest of the client can use the remaining four special streams using multi user memory and so some smart phones can still get some push messages which usually would not be sended yet but they can already be distributed that gives us some huge advantage like for example for latency in the entire network because smaller data transmission can just be a throne out there so the best thing is you can actually pick a different MCS index per user so we are sending and we have the choice for every different user to use a different modulation a different redundancy screen that's really a technical achievement where sometimes think how was this possible how did they implement this building something like that that's really something so then there was this other thing beam shaping that's really cool stuff really it's cool stuff for example in my hackerspace they now build a they build a speaker which uses beamforming of audio and it sends audio in just one direction so something you didn't hear I actually was somebody like has put the nestless song and you couldn't hear it because it was beam shaped only in my direction it's really just an active modulation of the dissipation behavior in high frequency range so in 8.2.11ac this gives us an additional gain in terms of signal of roundabout 2.5dB because we can push our data exactly into one direction and that's much better because the further we are away from the access point the worse our signal gets of course so we we are sliding into lower mcs indices and we can get less data so if we can now amplify our transmission into one direction that has the advantage that we can get more data pushed through this so we have the advantage that we are finished with our transmission earlier so everybody else also gets more atime to use this network beam shaping as all you already mentioned earlier so we had existed in 8.2.11.n but there were so many weird things there and now they finally agreed with 11ac and it's actually possible bi-directional there's hardly any clients supporting this because most clients only have 2 or 3 antennas for 2 or 3 special streams so there's not much support for this there but in particular in the enterprise area manufacturers started implementing this beam shaping and it worked really nicely but not for the uplink it doesn't always work for the uplink, it's not always possible so here I plotted this, I took the relative distance to the access point and plotted the mcs indices the lower arrow is what happens when we use beam shaping and the 2.5 db gain which we got and we can be way further away from the access point and we can use a higher mcs index so we can get the same data transmitted in a higher distance from the access point which gives us some compensation for the loss that 5GHz has anyway because of the higher damping now we're going to talk about phased array antennas phased array antennas is a space efficient way of replacing a Yagi antenna and if I have a very big antenna then I need a motor and it needs to start, it takes time and the phased array can very quickly change the direction and we can do this in this case it is technically it is technically very cumbersome but it is possible we do this through a phase shift during the transmission we have a number of antennas which are parallel to each other and if we start sending the signal on one end or the other ones then the whole signal changes direction to the side and through this mechanism the phased array is controlled you need to compute an individual phase per antenna you can implement its fix on a board so if you build a radar phased array in a car then you make the wire to the one antenna longer than to the other and that gives you a direction and here is a picture showing how it works this is the only picture I've taken from Wikipedia there are no pretty pictures to this topic otherwise so if one of you guys wants to if you want to load them into Wikipedia then you can talk to me later and I'll give them to you happily so they've decided on this thing, null data packet beamforming and it really requires that you measure the channel before you do your transmission then you have to distinguish between the one who forms the beam and and the client who receives the formed beam and the terms of beamformer and beamformy the angle of the transmission will be held in matrices matrices are better than angles we have two matrices we have the feedback matrix the one that we get back from the client on how he hears us and we have the steering matrix which gets applied to our transmission in order to form the transmission in the standard you can read it but it is really painful to read so the null data packet beamforming you start with an announcement that you want to measure then you start sending zero data packets and now the client knows what the packet is and he can measure the direction und und sich als feedback matrix ansprechen und dann senden sie zurück und dann die Master start sending the actual data und das data will dann be received by the client so since we know the IEEE are crazy we are not only doing beamforming with one client we are doing it with multiple clients so we are doing multi-user Mimo and we are doing beamforming to multiple clients at the same time so the technical implementation is very very good first he sends an announcement then he sends the packet and then he asks for the feedback measuring the channel does cost some airtime the size of the feedback matrix is different depending on the number of clients depending on the number of partial streams of the clients and some other parameters the width of the channel is part of the determining factors and this matrix can be anywhere from 78 bytes to 53 kilobytes also variiert sehr stark das ist eine große Variation in der Größe so 1,5% der Airtime ist für die Measurements of beamforming consumed wir sind surprisingly akkurat für jede Subcarrier sie können 56 Angles wenn sie 8 Partial Streams machen und die Richtung drücken und das ist eigentlich wenn man es mal genau überlegt und auch die Geschwindigkeit also wenn du das denkst und du denkst um die Speed in der das Data wird transmitt dann ist das ein sehr akkurat Targeting so jetzt wir werden die Realität in der Theorie und ich liebe das Standard und es ist fantastisch aber unfortunately die Datenrateen in der Realität sind schneller wenn du schaust wenn du die Akkurspunkte schaust hier du wirst nicht so viel Data durch als die Standard-Promissen da sind zwei Räume für das wir haben sehr viele Leute in der Raum eingegrenzt und wir haben restriktiert, wie weit der Channel kann sein ich habe vorher erklärt warum man die Channel-Size restrikt dann also er ist sehr unhappig 2,4 GHz-Kliens also er ist auch unhappig um 5 AC-Kliens 11 AC-Kliens um 5 GHz er ist auch unhappig weil sie auch nicht konformiert sind mit diesen Legacy-Kliens hier im Kongress gibt es einen guten Modell-Equipment 75% der Leute haben die Endstandard die Broadcast wird mit einem hohen Data-Rate also wenn du ein Klient und die Akkurspunkte negatieren dann wird der tiefe 1 oder 2 dann wird der andere 2 also der User 80 MHz 160 MHz-Channels ist sehr schwierig in Deutschland wir sind sehr limitiert zu einigen Channels die Bands sind sehr klein und in Deutschland können wir einfach 80 MHz-Channels finden in diesen Bands und das ist ein das dieser Standard nicht so vernünftig implementiert wurde so die Implementation ist nicht immer genau wie der Standard es proschribe also die Manufacturers haben etwas ermittelt sie haben nicht die beamformen unilaterale oder letztes Jahr rausgekommen sind die jetzt hier irgendwie die ganze Zeit rumhängen können das alle nicht die Akkurspunkte die wir hier haben und das ganze macht es dann natürlich auch mal ein bisschen schwieriger weil wir auch wieder Datenraten zurückfallen dann hat auf dieses Ausfall coole Ideen in addition this great idea of deployment things in terms of wave didn't really work properly but then people thought well wave 2 is what we have to support at least so that's good enough so far I have not found any client which actually supports 8 special streams no standard actually supports the entire standard I found a chipset but the chipset doesn't give me anything if I don't find a chip which I can put below this and then I can use this the problem with this in some extent is the power supply an access point needs power somehow obviously if we if we use power over ethernet or poe plus to do this for example following 802.11.380 that's roughly 25 Watt that's very cool that's enough usually however if we now start doing spatial mapping where we split the data stream onto the different spatial streams so that they can be put together back then that requires a big digital CPU which can process all of this the more streams we want to use in parallel the bigger the CPU has to be so the more power it is using so that's unfortunately still a problem getting the power to the AP so beamforming as already mentioned is not very widely not very widespread also the uplink of the access point is long limit of the standard most access points have one or two gigabit so the first standard that actually support 2.5 gigabit for the uplink but you don't need this at the university at our access points we see maximal uplink usage of 200 megabit also here in congress access point don't even remotely reach the limits of the standard I have not yet seen a single access point which actually was able to get the throughput of a gigabit from the Wi-Fi to the LAN that's probably possible in the lab but of course when you have a Wi-Fi and it's not about the lab it's about actually getting this in the wild where there's actual people so some people for example live in the city and not on the LAN on the countryside of course you have all the frequency for yourself but in the city you're sharing it with all your neighbors and things don't work as well I got some nice ideas all of this has a future things have to be optimized further IEEE is by a long reach not at the point yet where they are like that's how we want it, that's how we want to use it that's how everything will stay so the hunger for more data throughput is not yet satisfied we need better support for very high density deployments like for example in rooms like this where the access points and the clients we have some have to make sure that they don't disturb and annoy each other so much we have to make sure everything interacts with each other better so to this extent we have 802.11ax-2019 yes, that's the new thing if you think 11ac is sexy you didn't see this one it's again another huge step so far unfortunately I wasn't able to get draft 1.0 if somebody has access to the IEEE draft I would look at this I would really like to look at this, my university does get the standards but only the ones that are finished, not the drafts I would really like to have this I would really like to read it because just looking at papers you don't get the ideas with this one we get 1024 QAM two more MCS values much higher throughput so it's going to be awesome however with this standard it's not just about more data let's get some more data but let's also optimise some other things so this multi-user MIMO let's do this bi-directional so that multiple not only the access point can send multiple data streams to different clients that also different clients can send at the same time their independent data streams to the access point that's really cool but OFDMA which is for frequency directive multiple access that's a huge route it's really complicated most of you have it in your in your backstor, LTE is using that at the same time multiple users can use the different sub-carriers of a transmission and they get very weird nested access to the same channel I plan to do something about 802.11ax at the next GPN where I will explore this further I'm already almost using up my time here so with OFDMA everything will be much nicer and I'm really really looking forward to the standard there's already the first chips based on the draft 1.0 so hardware developers are now very much invited to implement the standard thank you very much and now I'm looking forward to your questions we got some time for Q&A if you want to leave please take your trash with you but we got some 10 minutes for Q&A the mic here in the hall question hello you mentioned that the matrices I think with the beamforming the matrices are different in size ist das because the matrices get more rows and columns or the answer is yes yes I think that's because they have to contain more data because from 8 special streams you have to specify more precisely what the angle is so the individual values have more data so that will be my question the values get larger the answer is both mic on the side sorry for asking in English question is actually in English if you take 8 special streams and you take your 360° antenna array which is placed in a circle you just divide your 360° through the 56° at your angle which you can reach with beamforming so another question from the internet for once lots of applause from the internet so now the question from the internet two questions summarised at one how does much movement of the clients how does this effect beamforming work and on the other hand is it possible to control this do you see more potential with beamforming to extend this answer yes I do see lots of potential with beamforming to extend this for example you could support more special streams but then you also need more power how does beamforming work when lots of people leave the room well when all of these people suddenly leave the room I am going to find you in most of the cases they won't just run around and transmit data all the time beamforming does cost lots of airtime however in principle beamforming is very quick it doesn't even take like a millisecond to measure and transmit this because the angles are really are slightly wider so it's still possible for the clients to move in this and well in doubt it's going to be a missed error in der Transmission so at that point it's really nice to have a ccp question disclaimer I am a software person so all this hardware is just voodoo for me so I wondered how do you measure these things how do you debug answer so what do you mean question everything measuring high frequency so my professor says when you ask him how do you measure frequency that's several lectures to explain this to develop this high frequency stuff you have to set it up in parts compute everything measure everything so as a user doing troubleshooting that's difficult you have to rely on the chips that were used there to actually work properly I know people that are now starting to look at the HTA 10K Binary Blob and reverse engineered to fix the bugs in it and to figure out what is going on there and how things are working if you don't actually have contact to the sources then troubleshooting is really hard question so what about this beamforming in this 807 standard I got a I'm not allowed to send stronger than 100mW so with beamforming I got a 2.5 dB increase in power is this okay legally if anybody would care answer technically speaking technically speaking not the access point would have to be careful however in this excitement another advantage with beamforming is if we have access points in one direction and one in another direction the access points are not disturbing as much anymore but looking at the legalies yes they would not be allowed to go beyond this in the transmission so if you have a question if you have a really annoyed neighbor they could sue you but they would have to measure this and also it would be the manufacturer's fault and not yours in your frequency plan you have channel 144 through 149 and what is the reason secondly in the NTP announcements are they sent isosynchronous or which times are they sent in is it done on demand or how is it done null datapacketforming works in such a way so that they send that they measure before every transmission because the clients aren't to be expected in the same place and when you are using beamforming then you are expecting multiple clients so it may be a very different client and so you measure again you were talking about the channels 149 the channels I am interested in 144 through 149 has a gap the channels exist theoretically but they were denied from the standardization the regulator said this cannot be used we don't know the reasons it is not a legacy reason it might be radar when I made the picture in the train I thought I have to explain it but I haven't so I can twitter it later okay last question thank you for the talk the raw data rate is one thing but is there anything else that has been improved with AC that you should talk about yes layer 2 of the standards has made changes and improvements but I would need to show you packets and explain why there is a 1 and a 0 and why there would be an improvement the difference of packets between 11n and 11ac would be too theoretical and boring for this talk here I can recommend some literature on the topic 802.11ac the guide the book is called 802.11ac the definite guide from Matthew Asgast and he explains how these things works and how the packets are structured okay so one more question and one from the internet you mentioned AX then I heard AD does one of those standards steal the airtime of the other? no it doesn't I have heard rohme in one of the standards yes there are rohme standards in 802.11ac I think there is about 3 of them yes there is also standardized ones it always takes time for the standards to become established and the manufacturers haven't really made hard for it yet the standards do not hurt anybody when they are not implemented but in some cases the rohme dies and that's why sometimes you have to use proprietary extensions that the other manufacturers haven't done right final question comes from the internet the question is do you sniff memo systems and don't you need the channel matrix and the answer is yeah it's a data transmission on layer one and you can sniff it you need the same hardware on the other side you would need to use a software defined radio and the beamforming matrix is not a problem either it isn't used to change data it's only to adjust the directionality beamforming is not so accurate that all data goes in just one direction if you have a good enough antenna you can still listen in thank you very much