From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp67.iad3a.emailsrvr.com (smtp67.iad3a.emailsrvr.com [173.203.187.67]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id 0D8C53CB45 for ; Fri, 3 Sep 2021 14:33:35 -0400 (EDT) Received: from app59.wa-webapps.iad3a (relay-webapps.rsapps.net [172.27.255.140]) by smtp1.relay.iad3a.emailsrvr.com (SMTP Server) with ESMTP id 43E0E443B; Fri, 3 Sep 2021 14:33:35 -0400 (EDT) Received: from deepplum.com (localhost.localdomain [127.0.0.1]) by app59.wa-webapps.iad3a (Postfix) with ESMTP id 2EAB960066; Fri, 3 Sep 2021 14:33:35 -0400 (EDT) Received: by apps.rackspace.com (Authenticated sender: dpreed@deepplum.com, from: dpreed@deepplum.com) with HTTP; Fri, 3 Sep 2021 14:33:35 -0400 (EDT) X-Auth-ID: dpreed@deepplum.com Date: Fri, 3 Sep 2021 14:33:35 -0400 (EDT) From: "David P. Reed" To: "Matt Mathis" Cc: dickroy@alum.mit.edu, "Cake List" , "Make-Wifi-fast" , "Bob McMahon" , starlink@lists.bufferbloat.net, "codel" , "cerowrt-devel" , "bloat" , "Rodney W. Grimes" MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_20210903143335000000_27193" Importance: Normal X-Priority: 3 (Normal) X-Type: html In-Reply-To: References: <202108101410.17AEAR4w075939@gndrsh.dnsmgr.net> <5AF5551E2A7041168E7071FDA0F6B8EC@SRA6> <03CA2CDA3EC5415DA229F835BE039994@SRA6> <1630604193.476312238@apps.rackspace.com> X-Client-IP: 209.6.168.128 Message-ID: <1630694015.155723881@apps.rackspace.com> X-Mailer: webmail/19.0.11-RC X-Classification-ID: 014209b0-b0fb-4c2c-ac4c-2041c6fe2a22-1-1 Subject: Re: [Cake] [Bloat] [Starlink] [Make-wifi-fast] [Cerowrt-devel] Due Aug 2: Internet Quality workshop CFP for the internet architecture board X-BeenThere: cake@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: Cake - FQ_codel the next generation List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Sep 2021 18:33:36 -0000 ------=_20210903143335000000_27193 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable =0ARegarding "only needs to be solved ... high density" - Musk has gone on = record as saying that Starlink probably will never support dense subscriber= areas. Which of course contradicts many other statements by Starlink and S= tarfans that they can scale up to full coverage of the world. My point in t= his regard is that "armchair theorizing" is not going to discover how scala= ble Starlink technology (or LEO technology) can be, because there are many,= many physical factors besides constellation size that will likely limit sc= aling.=0A =0AIt really does bug me that Musk and crew have promised very lo= w latency as a definite feature of Starlink, but then couldn't seem to even= bother to get congestion control in their early trial deployments.=0AThat = one should be solvable.=0A =0ABut they are declaring victory and claiming t= hey have solved every problem, so they should get FCC permission to roll ou= t more of their unproven technology, right now. Reminds me of ATT deploying= the iPhone. As soon as it stopped working very well after the early raving= reviews from early adopters, ATT's top technology guy (a John Donavan) wen= t on a full on rampage against Apple for having a "defective product" when = in fact it was ATT's HSPA network that was getting severely congested due t= o its extreme bufferbloat design. (It wasn't ATT, it was actually Alcatel L= ucent that did the terrible design, but ATT continued to blame Apple.)=0A = =0ASince some on this list want to believe that Starlink is the savior, but= others are technically wise, I'm not sure where the discussion will go. I = hope that there will be some feedback to Starlink rather than just a fan cl= ub or user-support group.=0A =0A =0AOn Friday, September 3, 2021 10:35am, "= Matt Mathis" said:=0A=0A=0A=0AI am very wary of a g= eneralization of this problem: software engineers who believe that they can= code around arbitrary idosynchronies of network hardware. They often succ= eed, but generally at a severe performance penalty.=0AHow much do we know a= bout the actual hardware? As far as I understand the math, some of the pr= ime calculations used in Machine Learning are isomorphic to multidimensiona= l correlators and convolutions, which are the same computations as needed t= o do phased array beam steering. One can imagine scenarios where Tesla (p= lans to) substantially overbuild the computational HW by recycling some ML = technology, and then beefing up the SW over time as they better understand = reality.=0AAlso note that the problem really only needs to be solved in are= as where they will eventually have high density. Most of the early deploy= ment will never have this problem.=0A=0A=0A=0A=0A=0A=0A=0A=0A=0AThanks,--MM= --=0AThe best way to predict the future is to create it. - Alan Kay=0A=0AW= e must not tolerate intolerance;=0A however our response must be care= fully measured: =0A too strong would be hypocritical and risks s= piraling out of control;=0A too weak risks being mistaken for ta= cit approval.=0A=0A=0AOn Thu, Sep 2, 2021 at 10:36 AM David P. Reed <[ dpre= ed@deepplum.com ]( mailto:dpreed@deepplum.com )> wrote:=0AI just want to th= ank Dick Roy for backing up the arguments I've been making about physical R= F communications for many years, and clarifying terminology here. I'm not t= he expert - Dick is an expert with real practical and theoretical experienc= e - but what I've found over the years is that many who consider themselves= "experts" say things that are actually nonsense about radio systems.=0A = =0AIt seems to me that Starlink is based on a propagation model that is qui= te simplistic, and probably far enough from correct that what seems "obviou= s" will turn out not to be true. That doesn't stop Musk and cronies from as= serting these things as absolute truths (backed by actual professors, espec= ially professors of Economics like Coase, but also CS professors, network p= rotocol experts, etc. who aren't physicists or practicing RF engineers).=0A= =0AThe fact is that we don't really know how to build a scalable LEO syste= m. Models can be useful, but a model can be a trap that causes even enginee= rs to be cocky. Or as the saying goes, a Clear View doesn't mean a Short Di= stance.=0A =0AIf there are 40 satellites serving 10,000 ground terminals si= multaneously, exactly what is the propagation environment like? I can tell = you one thing: if the phased array is digitized at some sample rate and som= e equalization and some quantization, the propagation REALLY matters in ser= ving those 10,000 ground terminals scattered randomly on terrain that is no= t optically flat and not fully absorbent.=0A =0ASo how will Starlink scale?= I think we literally don't know. And the modeling matters.=0A =0ARecently = a real propagation expert (Ted Rapaport and his students) did a study of ho= w well 70 GHz RF signals propagate in an urban environment - Brooklyn. The= standard model would say that coverage would be terrible! Why? Because sup= posedly 70 GHz is like visible light - line of sight is required or nothing= works.=0A =0ABut in fact, Ted, whom I've known from being on the FCC Techn= ological Advisory Committee (TAC) together when it was actually populated w= ith engineers and scientists, not lobbyists, discovered that scattering and= diffraction at 70 GHz in an urban environment significantly expands covera= ge of a single transmitter. Remarkably so. Enough that "cellular architectu= re" doesn't make sense in that propagation environment.=0A =0ASo all the pr= ofessional experts are starting from the wrong place, and amateurs perhaps = even more so.=0A =0AI hope Starlink views itself as a "research project". I= 'm afraid it doesn't - partly driven by Musk, but equally driven by the FCC= itself, which demands that before a system is deployed that the entire pla= n be shown to work (which would require a "model" that is actually unknowab= le because something like this has never been tried). This is a problem wit= h today's regulation of spectrum - experiments are barred, both by law, and= by competitors who can claim your system will destroy theirs and not work.= =0A =0ABut it is also a problem when "fans" start setting expectations way = too high. Like claiming that Starlink will eliminate any need for fiber. We= don't know that at all!=0A =0A =0A =0A =0A =0A =0A =0AOn Tuesday, August 1= 0, 2021 2:11pm, "Dick Roy" <[ dickroy@alum.mit.edu ]( mailto:dickroy@alum.m= it.edu )> said:=0A=0A=0A=0A=0ATo add a bit more, as is easily seen below, t= he amplitudes of each of the transfer functions between the three transmit = and three receive antennas are extremely similar. This is to be expected, = of course, since the =E2=80=9Caperture=E2=80=9D of each array is very small= compared to the distance between them. What is much more interesting and = revealing is the relative phases. Obviously this requires coherent receive= rs, and ultimately if you want to control the spatial distribution of power= (aka SDMA (or MIMO in some circles) coherent transmitters. It turns out th= at just knowing the amplitude of the transfer functions is not really all t= hat useful for anything other than detecting a broken solder joint:^)))=0A = =0AAlso, do not forget that depending how these experiments were conducted,= the estimates are either of the RF channel itself (aka path loss),or of th= e RF channel in combination with the transfer functions of the transmitters= and//or receivers. What this means is the CALIBRATION is CRUCIAL! Those = who do not calibrate, are doomed to fail!!!! I suspect that it is in cali= bration where the major difference in performance between vendors=E2=80=99= =E2=80=99 products can be found :^))))=0A =0AIt=E2=80=99s complicated =E2= =80=A6 =0A =0A=0A=0AFrom: Bob McMahon [mailto:[ bob.mcmahon@broadcom.com ](= mailto:bob.mcmahon@broadcom.com )] =0ASent: Tuesday, August 10, 2021 10:07= AM=0ATo: [ dickroy@alum.mit.edu ]( mailto:dickroy@alum.mit.edu )=0ACc: Rod= ney W. Grimes; Cake List; Make-Wifi-fast; [ starlink@lists.bufferbloat.net = ]( mailto:starlink@lists.bufferbloat.net ); codel; cerowrt-devel; bloat=0AS= ubject: Re: [Starlink] [Cake] [Make-wifi-fast] [Cerowrt-devel] Due Aug 2: I= nternet Quality workshop CFP for the internet architecture board=0A =0A=0AT= he slides show that for WiFi every transmission produces a complex frequenc= y response, aka the h-matrix. This is valid for that one transmission only.= The slides show an amplitude plot for a 3 radio device hence the 9 elemen= ts per the h-matrix. It's assumed that the WiFi STA/AP is stationary such t= hat doppler effects aren't a consideration. WiFi isn't a car trying to conn= ect to a cell tower. The plot doesn't show the phase effects but they are = included as the output of the channel estimate is a complex frequency respo= nse. Each RX produces the h-matrix ahead of the MAC. These may not be symme= tric in the real world but that's ok as transmission and reception is one w= ay only, i.e. the treating them as repcripocol and the matrix as hollows sy= mmetric isn't going to be a "test blocker" as the goal is to be able to use= software and programmable devices to change them in near real time. The cu= rrent approach used by many using butler matrices to produce off-diagonal e= ffects is woefully inadequate. And we're paying about $2.5K per each butle= r.=0A=0ABob=0A =0A=0A=0AOn Tue, Aug 10, 2021 at 9:13 AM Dick Roy <[ dickroy= @alum.mit.edu ]( mailto:dickroy@alum.mit.edu )> wrote:=0AWell, I hesitate t= o drag this out, however Maxwell's equations and the=0A invariance of the l= aws of physics ensure that all path loss matrices are=0A reciprocal. What = that means is that at any for any given set of fixed=0A boundary conditions= (nothing moving/changing!), the propagation loss between=0A any two points= in the domain is the same in both directions. The=0A "multipathing" in one= direction is the same in the other because the=0A two-parameter (angle1,an= gle2) scattering cross sections of all objects=0A (remember they are fixed = here) are independent of the ordering of the=0A angles. =0A=0A Very import= antly, path loss is NOT the same as the link loss (aka link=0A budget) whic= h involves tx power and rx noise figure (and in the case of=0A smart antenn= as, there is a link per spatial stream and how those links are=0A managed/c= ontrolled really matters, but let's just keep it simple for this=0A discuss= ion) and these generally are different on both ends of a link for a=0A vari= ety of reasons. The other very important issue is that of the=0A ""measurem= ent plane", or "where tx power and rx noise figure are being=0A measured/re= ferenced to and how well the interface at that plane is=0A "matched". We g= enerally assume that the matching is perfect, however it=0A never is. All o= f these effects contribute to the link loss which determines=0A the strengt= h of the signal coming out of the receiver (not the receive=0A antenna, the= receiver) for a given signal strength coming out of the=0A transmitter (no= t the transmit antenna, the tx output port). =0A=0A In the real world, th= ings change. Sources and sinks move as do many of the=0A objects around th= em. This creates a time-varying RF environment, and now=0A the path loss m= atrix is a function of time and a few others things, so it=0A matters WHEN = something is transmitted, and WHEN it is received, and the two=0A WHEN's ar= e generally separated by "the speed of light" which is a ft/ns=0A roughly. = As important is the fact that it's no longer really a path loss=0A matrix c= ontaining a single scalar because among other things, the time=0A varying e= nvironment induces change in the transmitted waveform on its way to=0A the = receiver most commonly referred to as the Doppler effect which means=0A the= re is a frequency translation/shift for each (multi-)path of which there=0A= are in general an uncountably infinite number because this is a continuous= =0A world in which we live (the space quantization experiment being conduct= ed in=0A the central US aside:^)). As a consequence of these physical laws,= the=0A entries in the path loss matrix become complex functions of a numbe= r of=0A variables including time. These functions are quite often character= ized in=0A terms of Doppler and delay-spread, terms used to describe in jus= t a few=0A parameters the amount of "distortion" a complex function causes.= =0A=0A Hope this helps ... probably a bit more than you really wanted to k= now as=0A queuing theorists, but ...=0A=0A -----Original Message-----=0A Fr= om: Starlink [mailto:[ starlink-bounces@lists.bufferbloat.net ]( mailto:sta= rlink-bounces@lists.bufferbloat.net )] On Behalf Of=0A Rodney W. Grimes=0A = Sent: Tuesday, August 10, 2021 7:10 AM=0A To: Bob McMahon=0A Cc: Cake List;= Make-Wifi-fast; [ starlink@lists.bufferbloat.net ]( mailto:starlink@lists.= bufferbloat.net );=0A[ codel@lists.bufferbloat.net ]( mailto:codel@lists.bu= fferbloat.net ); cerowrt-devel; bloat=0A Subject: Re: [Starlink] [Cake] [Ma= ke-wifi-fast] [Cerowrt-devel] Due Aug 2:=0A Internet Quality workshop CFP f= or the internet architecture board=0A=0A > The distance matrix defines sign= al attenuations/loss between pairs. It's=0A > straightforward to create a = distance matrix that has hidden nodes because=0A > all "signal loss" betwe= en pairs is defined. Let's say a 120dB=0A attenuation=0A > path will cause= a node to be hidden as an example.=0A > =0A > A B C D=0A > = A - 35 120 65=0A > B - 65 65=0A > C - = 65=0A > D -=0A > =0A > So in the above, AC ar= e hidden from each other but nobody else is. It does=0A > assume symmetry b= etween pairs but that's typically true.=0A=0A That is not correct, symmetry= in the RF world, especially wifi, is rare=0A due to topology issues. A hi= gh transmitter, A, and a low receiver, B,=0A has a good path A - > B, but = a very weak path B -> A. Multipathing=0A is another major issue that caus= es assymtry.=0A=0A > =0A > The RF device takes these distance matrices as s= ettings and calculates the=0A > five branch tree values (as demonstrated in= the video). There are=0A > limitations to solutions though but I've found = those not to be an issue to=0A > date. I've been able to produce hidden nod= es quite readily. Add the phase=0A > shifters and spatial stream powers can= also be affected, but this isn't=0A > shown in this simple example.=0A > = =0A > Bob=0A > =0A > On Mon, Aug 2, 2021 at 8:12 PM David Lang <[ david@lan= g.hm ]( mailto:david@lang.hm )> wrote:=0A > =0A > > I guess it depends on w= hat you are intending to test. If you are not=0A going=0A > > to=0A > > tin= ker with any of the over-the-air settings (including the number of=0A > > p= ackets=0A > > transmitted in one aggregate), the details of what happen ove= r the air=0A > > don't=0A > > matter much.=0A > >=0A > > But if you are goi= ng to be doing any tinkering with what is getting=0A sent,=0A > > and=0A > = > you ignore the hidden transmitter type problems, you will create a=0A > >= solution that=0A > > seems to work really well in the lab and falls on it'= s face out in the=0A > > wild=0A > > where spectrum overload and hidden tra= nsmitters are the norm (at least=0A in=0A > > urban=0A > > areas), not rare= corner cases.=0A > >=0A > > you don't need to include them in every test, = but you need to have a way=0A > > to=0A > > configure your lab to include t= hem before you consider any=0A > > settings/algorithm=0A > > ready to try i= n the wild.=0A > >=0A > > David Lang=0A > >=0A > > On Mon, 2 Aug 2021, Bob = McMahon wrote:=0A > >=0A > > > We find four nodes, a primary BSS and an adj= unct one quite good for=0A lots=0A > > of=0A > > > testing. The six nodes = allows for a primary BSS and two adjacent=0A ones.=0A > > We=0A > > > want = to minimize complexity to necessary and sufficient.=0A > > >=0A > > > The c= hallenge we find is having variability (e.g. montecarlos) that's=0A > > > r= eproducible and has relevant information. Basically, the distance=0A > > ma= trices=0A > > > have h-matrices as their elements. Our chips can provide th= ese=0A > > h-matrices.=0A > > >=0A > > > The parts for solid state programm= able attenuators and phase shifters=0A > > > aren't very expensive. A devic= e that supports a five branch tree and=0A 2x2=0A > > > MIMO seems a very go= od starting point.=0A > > >=0A > > > Bob=0A > > >=0A > > > On Mon, Aug 2, 2= 021 at 4:55 PM Ben Greear <[ greearb@candelatech.com ]( mailto:greearb@cand= elatech.com )>=0A > > wrote:=0A > > >=0A > > >> On 8/2/21 4:16 PM, David La= ng wrote:=0A > > >>> If you are going to setup a test environment for wifi,= you need to=0A > > >> include the ability to make a fe cases that only hap= pen with RF, not=0A > > with=0A > > >> wired networks and=0A > > >>> are co= mmonly overlooked=0A > > >>>=0A > > >>> 1. station A can hear station B and= C but they cannot hear each=0A other=0A > > >>> 2. station A can hear stat= ion B but station B cannot hear station A=0A 3.=0A > > >> station A can hea= r that station B is transmitting, but not with a=0A strong=0A > > >> enough= signal to=0A > > >>> decode the signal (yes in theory you can work around = interference,=0A but=0A > > >> in practice interference is still a real thi= ng)=0A > > >>>=0A > > >>> David Lang=0A > > >>>=0A > > >>=0A > > >> To add = to this, I think you need lots of different station devices,=0A > > >> diff= erent capabilities (/n, /ac, /ax, etc)=0A > > >> different numbers of spati= al streams, and different distances from=0A the=0A > > >> AP. From downloa= d queueing perspective, changing=0A > > >> the capabilities may be sufficie= nt while keeping all stations at same=0A > > >> distance. This assumes you= are not=0A > > >> actually testing the wifi rate-ctrl alg. itself, so diff= erent=0A throughput=0A > > >> levels for different stations would be enough= .=0A > > >>=0A > > >> So, a good station emulator setup (and/or pile of rea= l stations) and=0A a=0A > > few=0A > > >> RF chambers and=0A > > >> program= mable attenuators and you can test that setup...=0A > > >>=0A > > >> From = upload perspective, I guess same setup would do the job.=0A > > >> Queuing/= fairness might depend a bit more on the=0A > > >> station devices, emulated= or otherwise, but I guess a clever AP could=0A > > >> enforce fairness in = upstream direction=0A > > >> too by implementing per-sta queues.=0A > > >>= =0A > > >> Thanks,=0A > > >> Ben=0A > > >>=0A > > >> --=0A > > >> Ben Greea= r <[ greearb@candelatech.com ]( mailto:greearb@candelatech.com )>=0A > > >>= Candela Technologies Inc [ http://www.candelatech.com ]( http://www.cande= latech.com )=0A > > >>=0A > > >=0A > > >=0A > >=0A > =0A > -- =0A > This el= ectronic communication and the information and any files=0A transmitted =0A= > with it, or attached to it, are confidential and are intended solely for= =0A > the use of the individual or entity to whom it is addressed and may= =0A contain =0A > information that is confidential, legally privileged, pro= tected by privacy=0A=0A > laws, or otherwise restricted from disclosure to = anyone else. If you are =0A > not the intended recipient or the person resp= onsible for delivering the =0A > e-mail to the intended recipient, you are = hereby notified that any use, =0A > copying, distributing, dissemination, f= orwarding, printing, or copying of =0A > this e-mail is strictly prohibited= . If you received this e-mail in error, =0A > please return the e-mail to t= he sender, delete it from your computer, and =0A > destroy any printed copy= of it.=0A=0A [ Charset UTF-8 unsupported, converting... ]=0A > ___________= ____________________________________=0A > Starlink mailing list=0A > [ Star= link@lists.bufferbloat.net ]( mailto:Starlink@lists.bufferbloat.net )=0A > = [ https://lists.bufferbloat.net/listinfo/starlink ]( https://lists.bufferbl= oat.net/listinfo/starlink )=0A > =0A ______________________________________= _________=0A Starlink mailing list=0A[ Starlink@lists.bufferbloat.net ]( ma= ilto:Starlink@lists.bufferbloat.net )=0A[ https://lists.bufferbloat.net/lis= tinfo/starlink ]( https://lists.bufferbloat.net/listinfo/starlink )=0A=0ATh= is electronic communication and the information and any files transmitted w= ith it, or attached to it, are confidential and are intended solely for the= use of the individual or entity to whom it is addressed and may contain in= formation that is confidential, legally privileged, protected by privacy la= ws, or otherwise restricted from disclosure to anyone else. If you are not = the intended recipient or the person responsible for delivering the e-mail = to the intended recipient, you are hereby notified that any use, copying, d= istributing, dissemination, forwarding, printing, or copying of this e-mail= is strictly prohibited. If you received this e-mail in error, please retur= n the e-mail to the sender, delete it from your computer, and destroy any p= rinted copy of it._______________________________________________=0A Bloat = mailing list=0A[ Bloat@lists.bufferbloat.net ]( mailto:Bloat@lists.bufferbl= oat.net )=0A[ https://lists.bufferbloat.net/listinfo/bloat ]( https://lists= .bufferbloat.net/listinfo/bloat ) ------=_20210903143335000000_27193 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable

Regarding "only needs = to be solved ... high density" - Musk has gone on record as saying that Sta= rlink probably will never support dense subscriber areas. Which of course c= ontradicts many other statements by Starlink and Starfans that they can sca= le up to full coverage of the world. My point in this regard is that "armch= air theorizing" is not going to discover how scalable Starlink technology (= or LEO technology) can be, because there are many, many physical factors be= sides constellation size that will likely limit scaling.

=0A

 

=0A

It really does bug me that Musk = and crew have promised very low latency as a definite feature of Starlink, = but then couldn't seem to even bother to get congestion control in their ea= rly trial deployments.

=0A

That one should be solvab= le.

=0A

 

=0A

But they a= re declaring victory and claiming they have solved every problem, so they s= hould get FCC permission to roll out more of their unproven technology, rig= ht now. Reminds me of ATT deploying the iPhone. As soon as it stopped worki= ng very well after the early raving reviews from early adopters, ATT's top = technology guy (a John Donavan) went on a full on rampage against Apple for= having a "defective product" when in fact it was ATT's HSPA network that w= as getting severely congested due to its extreme bufferbloat design. (It wa= sn't ATT, it was actually Alcatel Lucent that did the terrible design, but = ATT continued to blame Apple.)

=0A

 

=0A

Since some on this list want to believe that Starlink is t= he savior, but others are technically wise, I'm not sure where the discussi= on will go. I hope that there will be some feedback to Starlink rather than= just a fan club or user-support group.

=0A

 =0A

 

=0A

On Friday, Sept= ember 3, 2021 10:35am, "Matt Mathis" <mattmathis@google.com> said:

=0A
=0A
I am v= ery wary of a generalization of this problem: software engineers who b= elieve that they can code around arbitrary idosynchronies of network hardwa= re.  They often succeed, but generally at a severe performance pe= nalty.=0A
How much do we know about the actual hardware?   As= far as I understand the math, some of the prime calculations use= d in Machine Learning are isomorphic to multidimensional correlators a= nd convolutions, which are the same computations as needed to do phase= d array beam steering.   One can imagine scenarios where Tesla (p= lans to) substantially overbuild the computational HW by recycling som= e ML technology, and then beefing up the SW over time as they better unders= tand reality.
=0A
Also note that the problem really only needs to = be solved in areas where they will eventually have high density.  &nbs= p;Most of the early deployment will never have this problem.
=0A<= div>=0A
=0A
=0A
=0A
=0A
=0A
= =0A
=0A
=0A
Thanks,
=0A--MM--
The best wa= y to predict the future is to create it.  - Alan Kay

We mus= t not tolerate intolerance;
=0A
      &= nbsp;however our response must be carefully measured: 
=0A
&n= bsp;           too strong would be hypocritical an= d risks spiraling out of control;
=0A
        =     too weak risks being mistaken for tacit approval.
=0A=0A
=0A
=0A
=0A
=0A
=0A
=0A
=0A
= =0A
=0A
=0A
On Thu, Sep 2, 2021 at 10:36 AM David P. Reed <dpreed@deepplum.com> wrote:
=0A
=0A

I just want to th= ank Dick Roy for backing up the arguments I've been making about physical R= F communications for many years, and clarifying terminology here. I'm not t= he expert - Dick is an expert with real practical and theoretical experienc= e - but what I've found over the years is that many who consider themselves= "experts" say things that are actually nonsense about radio systems.

= =0A

 

=0A

It seems to me th= at Starlink is based on a propagation model that is quite simplistic, and p= robably far enough from correct that what seems "obvious" will turn out not= to be true. That doesn't stop Musk and cronies from asserting these things= as absolute truths (backed by actual professors, especially professors of = Economics like Coase, but also CS professors, network protocol experts, etc= . who aren't physicists or practicing RF engineers).

=0A

 

=0A

The fact is that we don't really kno= w how to build a scalable LEO system. Models can be useful, but a model can= be a trap that causes even engineers to be cocky. Or as the saying goes, a= Clear View doesn't mean a Short Distance.

=0A

 = ;

=0A

If there are 40 satellites serving 10,000 grou= nd terminals simultaneously, exactly what is the propagation environment li= ke? I can tell you one thing: if the phased array is digitized at some samp= le rate and some equalization and some quantization, the propagation REALLY= matters in serving those 10,000 ground terminals scattered randomly on ter= rain that is not optically flat and not fully absorbent.

=0A

 

=0A

So how will Starlink scale? I th= ink we literally don't know. And the modeling matters.

=0A

 

=0A

Recently a real propagation expert= (Ted Rapaport and his students) did a study of how well 70 GHz RF signals = propagate in an urban environment - Brooklyn.  The standard model woul= d say that coverage would be terrible! Why? Because supposedly 70 GHz is li= ke visible light - line of sight is required or nothing works.

=0A

 

=0A

But in fact, Ted, whom I'v= e known from being on the FCC Technological Advisory Committee (TAC) togeth= er when it was actually populated with engineers and scientists, not lobbyi= sts, discovered that scattering and diffraction at 70 GHz in an urban envir= onment significantly expands coverage of a single transmitter. Remarkably s= o. Enough that "cellular architecture" doesn't make sense in that propagati= on environment.

=0A

 

=0A

 

=0A

I hope Starlink views itself as a "research project". I'm af= raid it doesn't - partly driven by Musk, but equally driven by the FCC itse= lf, which demands that before a system is deployed that the entire plan be = shown to work (which would require a "model" that is actually unknowable be= cause something like this has never been tried). This is a problem with tod= ay's regulation of spectrum - experiments are barred, both by law, and by c= ompetitors who can claim your system will destroy theirs and not work.

= =0A

 

=0A

But it is also a = problem when "fans" start setting expectations way too high. Like claiming = that Starlink will eliminate any need for fiber. We don't know that at all!=

=0A

 

=0A

 

=0A=

 

=0A

 

=0A

 

=0A

 

=0A

 

=0A

On Tuesday, August 10, 2021 2:11pm= , "Dick Roy" <= dickroy@alum.mit.edu> said:

=0A
=0A
=0A

T= o add a bit more, as is easily seen below, the amplitudes of each of the tr= ansfer functions between the three transmit and three receive antennas are = extremely similar.  This is to be expected, of course, since the =E2= =80=9Caperture=E2=80=9D of each array is very small compared to the distanc= e between them.  What is much more interesting and revealing is the re= lative phases.  Obviously this requires coherent receivers, and ultima= tely if you want to control the spatial distribution of power (aka SDMA (or= MIMO in some circles) coherent transmitters. It turns out that just knowin= g the amplitude of the transfer functions is not really all that useful for= anything other than detecting a broken solder joint:^)))

= =0A

 

=0A

Also, do not forget that depending how these experiments were conducted,= the estimates are either of the RF channel itself (aka path loss),or of th= e RF channel in combination with the transfer functions of the transmitters= and//or receivers.  What this means is the CALIBRATION is CRUCIAL!&nb= sp; Those who do not calibrate, are doomed to fail!!!!   I suspec= t that it is in calibration where the major difference in performance betwe= en vendors=E2=80=99=E2=80=99 products can be found :^))))

= =0A

 

=0A

It=E2=80=99s complicated =E2=80=A6

=0A

 

=0A
=0A
=0A

From: Bob McMahon [mailto:bob.mcmahon@broadcom.com]
Sent: Tuesday, August 1= 0, 2021 10:07 AM
To:= dickroy= @alum.mit.edu
Cc: Rodney W. Grimes; Cake List; Make-Wifi-fast; starlink@lists.bufferbloat= .net; codel; cerowrt-devel; bloat
Subject: Re: [Starlink] [Cake] [Make-wifi-fast] = [Cerowrt-devel] Due Aug 2: Internet Quality workshop CFP for the internet a= rchitecture board

=0A
=0A

 

=0A
=0AThe slides show that for WiFi every tra= nsmission produces a complex frequency response, aka the h-matrix. Thi= s is valid for that one transmission only.  The slides show an amplitu= de plot for a 3 radio device hence the 9 elements per the h-matrix. It's as= sumed that the WiFi STA/AP is stationary such that doppler effects aren't a= consideration. WiFi isn't a car trying to connect to a cell tower.  T= he plot doesn't show the phase effects but they are included as the output = of the channel estimate is a complex frequency response. Each RX produces t= he h-matrix ahead of the MAC. These may not be symmetric in the real world = but that's ok as transmission and reception is one way only, i.e. the = treating them as repcripocol and the matrix as hollows symmetric isn't goin= g to be a "test blocker" as the goal is to be able to use software and prog= rammable devices to change them in near real time. The current approach use= d by many using butler matrices to produce off-diagonal effects  = is woefully inadequate. And we're paying about $2.5K per each butler.
= 3D""
Bob

=0A
=0A

 

=0A
= =0A
=0A

dickroy@alum.mit.edu> wrote:

=0A
=0A
=0A

Well, I hesitate to drag this out, however Maxwell's equation= s and the
invariance of the laws of physics ensure that all path loss= matrices are
reciprocal.  What that means is that at any for an= y given set of fixed
boundary conditions (nothing moving/changing!), = the propagation loss between
any two points in the domain is the same= in both directions. The
"multipathing" in one direction is the same = in the other because the
two-parameter (angle1,angle2) scattering cro= ss sections of all objects
(remember they are fixed here) are indepen= dent of the ordering of the
angles. 

Very important= ly, path loss is NOT the same as the link loss (aka link
budget) whic= h involves tx power and rx noise figure (and in the case of
smart ant= ennas, there is a link per spatial stream and how those links are
man= aged/controlled really matters, but let's just keep it simple for this
discussion) and these generally are different on both ends of a link for = a
variety of reasons. The other very important issue is that of the ""measurement plane", or "where tx power and rx noise figure are being=
measured/referenced to and how well the interface at that plane is "matched".  We generally assume that the matching is perfect, how= ever it
never is. All of these effects contribute to the link loss wh= ich determines
the strength of the signal coming out of the receiver = (not the receive
antenna, the receiver) for a given signal strength c= oming out of the
transmitter (not the transmit antenna, the tx output= port).   

In the real world, things change.  So= urces and sinks move as do many of the
objects around them.  Thi= s creates a time-varying RF environment, and now
the path loss matrix= is a function of time and a few others things, so it
matters WHEN so= mething is transmitted, and WHEN it is received, and the two
WHEN's a= re generally separated by "the speed of light" which is a ft/ns
rough= ly. As important is the fact that it's no longer really a path loss
m= atrix containing a single scalar because among other things, the time
= varying environment induces change in the transmitted waveform on its way = to
the receiver most commonly referred to as the Doppler effect which= means
there is a frequency translation/shift for each (multi-)path o= f which there
are in general an uncountably infinite number because t= his is a continuous
world in which we live (the space quantization ex= periment being conducted in
the central US aside:^)). As a consequenc= e of these physical laws, the
entries in the path loss matrix become = complex functions of a number of
variables including time. These func= tions are quite often characterized in
terms of Doppler and delay-spr= ead, terms used to describe in just a few
parameters the amount of "d= istortion" a complex function causes.

Hope this helps ... prob= ably a bit more than you really wanted to know as
queuing theorists, = but ...

-----Original Message-----
From: Starlink [mailto= :starlink-bounces@lists.bufferbloat.net] On Behalf Of
Rodney W. = Grimes
Sent: Tuesday, August 10, 2021 7:10 AM
To: Bob McMahon Cc: Cake List; Make-Wifi-fast; starlink@lists.bufferbloat.net;
codel@lists.buf= ferbloat.net; cerowrt-devel; bloat
Subject: Re: [Starlink] [Cake]= [Make-wifi-fast] [Cerowrt-devel] Due Aug 2:
Internet Quality worksho= p CFP for the internet architecture board

> The distance mat= rix defines signal attenuations/loss between pairs.  It's
> s= traightforward to create a distance matrix that has hidden nodes because > all "signal  loss" between pairs is defined.  Let's say = a 120dB
attenuation
> path will cause a node to be hidden as= an example.
>
>      A    B&nb= sp;    C    D
> A   -   35=    120   65
> B        &nb= sp;-      65   65
> C     =          -       65
>= ; D                    &n= bsp;    -
>
> So in the above, AC are hidden f= rom each other but nobody else is. It does
> assume symmetry betwe= en pairs but that's typically true.

That is not correct, symmet= ry in the RF world, especially wifi, is rare
due to topology issues.&= nbsp; A high transmitter, A,  and a low receiver, B,
has a good = path A - > B, but a very weak path B -> A.   Multipathing is another major issue that causes assymtry.

>
&= gt; The RF device takes these distance matrices as settings and calculates = the
> five branch tree values (as demonstrated in the video). Ther= e are
> limitations to solutions though but I've found those not t= o be an issue to
> date. I've been able to produce hidden nodes qu= ite readily. Add the phase
> shifters and spatial stream powers ca= n also be affected, but this isn't
> shown in this simple example.=
>
> Bob
>
> On Mon, Aug 2, 2021 at 8= :12 PM David Lang <da= vid@lang.hm> wrote:
>
> > I guess it depends o= n what you are intending to test. If you are not
going
> >= ; to
> > tinker with any of the over-the-air settings (includin= g the number of
> > packets
> > transmitted in one = aggregate), the details of what happen over the air
> > don't > > matter much.
> >
> > But if you are = going to be doing any tinkering with what is getting
sent,
>= > and
> > you ignore the hidden transmitter type problems, = you will create a
> > solution that
> > seems to wo= rk really well in the lab and falls on it's face out in the
> >= wild
> > where spectrum overload and hidden transmitters are t= he norm (at least
in
> > urban
> > areas), no= t rare corner cases.
> >
> > you don't need to incl= ude them in every test, but you need to have a way
> > to
= > > configure your lab to include them before you consider any
= > > settings/algorithm
> > ready to try in the wild.
> >
> > David Lang
> >
> > On = Mon, 2 Aug 2021, Bob McMahon wrote:
> >
> > > We= find four nodes, a primary BSS and an adjunct one quite good for
lot= s
> > of
> > > testing.  The six nodes allo= ws for a primary BSS and two adjacent
ones.
> > We
= > > > want to minimize complexity to necessary and sufficient.
> > >
> > > The challenge we find is having vari= ability (e.g. montecarlos) that's
> > > reproducible and has= relevant information. Basically, the distance
> > matrices
> > > have h-matrices as their elements. Our chips can provide = these
> > h-matrices.
> > >
> > >= The parts for solid state programmable attenuators and phase shifters
> > > aren't very expensive. A device that supports a five branc= h tree and
2x2
> > > MIMO seems a very good starting p= oint.
> > >
> > > Bob
> > > > > > On Mon, Aug 2, 2021 at 4:55 PM Ben Greear <greearb@candelatech.com<= /a>>
> > wrote:
> > >
> > >>= ; On 8/2/21 4:16 PM, David Lang wrote:
> > >>> If you = are going to setup a test environment for wifi, you need to
> >= >> include the ability to make a fe cases that only happen with RF, = not
> > with
> > >> wired networks and
= > > >>> are commonly overlooked
> > >>>=
> > >>> 1. station A can hear station B and C but the= y cannot hear each
other
> > >>> 2. station A ca= n hear station B but station B cannot hear station A
3.
> &g= t; >> station A can hear that station B is transmitting, but not with= a
strong
> > >> enough signal to
> > &= gt;>> decode the signal (yes in theory you can work around interferen= ce,
but
> > >> in practice interference is still a = real thing)
> > >>>
> > >>> David= Lang
> > >>>
> > >>
> >= >> To add to this, I think you need lots of different station device= s,
> > >> different capabilities (/n, /ac, /ax, etc)
> > >> different numbers of spatial streams, and different di= stances from
the
> > >> AP.  From download que= ueing perspective, changing
> > >> the capabilities may b= e sufficient while keeping all stations at same
> > >> di= stance.  This assumes you are not
> > >> actually te= sting the wifi rate-ctrl alg. itself, so different
throughput
&= gt; > >> levels for different stations would be enough.
>= > >>
> > >> So, a good station emulator setup (= and/or pile of real stations) and
a
> > few
> &g= t; >> RF chambers and
> > >> programmable attenuato= rs and you can test that setup...
> > >>
> > = >>  From upload perspective, I guess same setup would do the job= .
> > >> Queuing/fairness might depend a bit more on the<= br /> > > >> station devices, emulated or otherwise, but I gues= s a clever AP could
> > >> enforce fairness in upstream d= irection
> > >> too by implementing per-sta queues.
= > > >>
> > >> Thanks,
> > >&g= t; Ben
> > >>
> > >> --
> >= >> Ben Greear <greearb@candelatech.com>
> > >> Candel= a Technologies Inc  http://www.candelatech.com
> > >>
> &= gt; >
> > >
> >
>
> -- > This electronic communication and the information and any files transmitted
> with it, or attached to it, are confidential an= d are intended solely for
> the use of the individual or entity t= o whom it is addressed and may
contain
> information that i= s confidential, legally privileged, protected by privacy

> l= aws, or otherwise restricted from disclosure to anyone else. If you are > not the intended recipient or the person responsible for deliverin= g the
> e-mail to the intended recipient, you are hereby notified= that any use,
> copying, distributing, dissemination, forwarding= , printing, or copying of
> this e-mail is strictly prohibited. I= f you received this e-mail in error,
> please return the e-mail t= o the sender, delete it from your computer, and
> destroy any pri= nted copy of it.

[ Charset UTF-8 unsupported, converting... ] > _______________________________________________
> Starli= nk mailing list
> Starlink@lists.bufferbloat.net
> http= s://lists.bufferbloat.net/listinfo/starlink
>
_________= ______________________________________
Starlink mailing list
Starlink@l= ists.bufferbloat.net
https://lists.bufferbloat.net/listinfo/sta= rlink

=0A
=0A
=0A


Bloat@lists.bufferblo= at.net
https://lists.bufferbloat.net/listinfo/b= loat

=0A
=0A
 ------=_20210903143335000000_27193--