Interesting People mailing list archives
more on Response to David Reed and Brett and the FRC
From: David Farber <dave () farber net>
Date: Fri, 15 Sep 2006 00:53:48 +0900
Begin forwarded message: From: "David P. Reed" <dpreed () reed com> Date: September 13, 2006 2:25:54 AM JST To: dave () farber net Cc: ip () v2 listbox com Subject: Re: [IP] more on Response to David Reed and Brett and the FRCDave - though I could continue my discussion/argument with Brett, I think it would be a waste of your readers' time. His response to my response continues to show his relatively weak grasp of the basics of electromagnetic waves, information theory, and the limits of current radio technology. There is no need for me to point those out to people who understand such things.
I do want to respond to his misreading of two of my statements, in order again to twist their meaning.
First: when I said on this list ""The Shannon-Hartley Theorem does not provide a limit to the communications capacity of the electromagnetic field...." I was making a very precise statement. The theorem does not. Brett's claim was that it did. Any physical limits arise from physical circumstances, in other words, from the physical constraints of the universe. SHT merely assumes that the physical environment of its channel consists of Gaussian White Noise. That is, that there is a probability distribution that predicts a process by which noise arrives at the receiver, which satisfies a particular model. The SHT does NOT assert that the world is built that way, and as I have said, physicists do not see that when they measure the world in experiments. Some very narrow experiments of very narrow bands in very isolated circumstances see GWN-like phenomena. But you really have to throw out datapoints and pretend that you are in a free, equilibrium vacuum (or construct a test chamber) to see GWN.
And of course, the "Shannon Limit" is just a relationship between signal power and achievable information rate given as a "known" the variance of Gaussian White Noise in its abstract channel (variance for Gaussian signals = power).
Second: he quotes from an interview "There's no scarcity of spectrum any more than there's a scarcity of the color green. We could instantly hook up to the Internet everyone who can pick up a radio signal, and they could pump through as many bits as they could ever want. We'd go from an economy of digital scarcity to an economy of digital abundance."
I speak and write as precisely as I can, and this quote was no exception. I stated "no scarcity". Brett would like that to be read as if I said there is "no limit to capacity". But what I was saying is precisely that there is no current scarcity of spectrum for all communications that would prevent the entire Internet from becoming wireless. Of course, the current form of regulation *does* prevent that very directly, by allocating frequencies to transmitters by service. This quote had nothing to do with the ShannonHartley theorem. It was referring to the idea of "interference" which says that two transmissions cannot be allowed to be in the same band at the same time, because information is "destroyed". That notion of "interference" is incorrect. Information is destroyed only by very complex, energetic suppression, in a system as linear as the electromagnetic field. What happens when you have multiple signals from multiple transmitters is superposition. Superposition leads to ambiguity when (and ONLY when) there is sufficiently unpredictable noise and insufficient sampling by the receiver. Increase the sampling, decrease the noise's predictability, and the ambiguity can be resolved.
It's not the quantity of "spectrum" that limits communication. Instead it is the design of the receiver system (including its antenna), the characteristics of the noise process (perhaps best described in terms of noise entropy, or information content, of the field on which signals are superposed), and the computational capabilities of the receiver (analog or digital) that are available to disambiguate the signals.
David Farber wrote:
Begin forwarded message: From: Brett Glass <brett () lariat net> Date: September 12, 2006 1:16:15 AM JST To: dave () farber net, ip () v2 listbox com Cc: Bob Frankston <bob () frankston com> Subject: Re: [IP] more on Response to David Reed and Brett and the FRC Bob Frankston writes:Our understanding and technology has come a long way since then and examples like the 25 Watt Voyager which is able to send kilobits of data per second against the background noise of Jupiter and now all the way from the Kuipers Belt demonstrate the spectrum allocation isn’t the only coding system and, in fact, it would be difficult to find one that is more wasteful. It’s analogous to allocating cattle ranches into quarter acre plots to avoid risking any bovine getting more than its “fair” share.It is not the allocation scheme, but rather the lack of will (and the lack of well developed technology) to enforce fair sharing of the plots that is the problem. To use a real estate analogy: There's no problem with setting aside some plots of land for exclusive private use and others for roads. But if there are no rules of the road, it's natural for everyone to want to use private roads because there is nothing to preclude antisocial behavior, malicious blocking of the roads, hogging of the roads, and/or other tragedies of the commons. Alas, in the world of spectrum, one of the reasons that good mechanisms for sharing have not been developed and deployed is that the private spectrum owners ("spectrum barons?") have actually discouraged it so as to make their private turf more valuable.Even if one accepts the need for spectrum allocation why do we go to the extreme of having to prove no possible interference to the feeblest of radios the analog ones that can’t get a clear signal anyway?It is true that there is an odd double standard. Radio and TV -- the media of the past -- are so sacrosanct that nothing may interfere with them. But wireless broadband services such as ours, which the public would dearly love to replace them, enjoy no such protection. And licensed spectrum has been placed beyond our grasp.I know Brett is using packet radios as a bypass and he depends on relatively long distance connections compared with in-home use.We are not merely using radios as a "bypass." Our broadband reaches areas which the cable/telco duopoly has no financial motivation to serve -- and provides higher throughput than them in any area we cover. We're not just a workaround; we're a first class communications medium and deserve consideration as such.The problems of using single frequency non-redundant signaling is exacerbated by channeling them into assigned bands. That alone requires a system for policing its use. The problems are deeper --the frequency regimen is fragile because the messages are not self- identifying so you must depend on others policing the bands on yourbehalf rather than taking responsibility for the relationships. There are many other consequences of this system any change requires the intervention of congress and decades to effect.In my ISART paper (cited previously), I've already mentioned these drawbacks of allocation by frequency. It's easy to demonstrate that it would be in everyone's best interest to assign subsets of a different and more flexible "signal space" instead. But even if we continue toallocate by frequency, we can certainly implement much better policies.Current policy is crippling not only my business (which I never expected to do as well as it has, considering the obstacles that are being erected in its path) but countless other existing and potential businesses. David Reed writes:I don't think Brett's comments read on anything I said. The major points I made were: 1) The Shannon-Hartley theorem does not define a limit to the capacity of the spectrum, which most definitely is what Brett was asserting.As I mentioned in my earlier message, "the spectrum" does not have a capacity. Communications links, or "channels" (as Shannon calls them), do. The way we allocate spectrum (a policy issue) and implement radios (a technical issue) determines how many clear, high capacity channels we can create. Both engineering and policy are important to making effective use of the available spectrum. That's why the FCC (a policy making body) has an Office of Engineering Technology. It is a shame that, too often, the OET's recommendations are brushed aside in favor of the desires of politically powerful corporations.2) Brett's comments were properly read as political,Actually, my comments were (and are) mostly related to science and engineering. The policy recommendations which I've included merely follow from them.I'm not sure where Brett concludes that I suggested that sophisticated technology could remove all constraints on electromagnetic communications.You wrote: "The Shannon-Hartley Theorem does not provide a limit to the communications capacity of the electromagnetic field...." You have also asserted that "Interference is a metaphor that paints an old limitation of technology as a fact of nature.... There's no scarcity of spectrumany more than there's a scarcity of the color green. We could instantlyhook up to the Internet everyone who can pick up a radio signal, and they could pump through as many bits as they could ever want." (See http://dir.salon.com/story/tech/feature/2003/03/12/spectrum/index.html for the full text of the article.) There and elsewhere, you have indeed claimed that sophisticated technology could remove all practical constraints on electromagnetic communications.You can read textbooks on electromagnetic physics at any level, and you will not find Shannon-Hartley.This is correct. This is because the Shannon-Hartley theorem relates to information theory and communications. Communications can be done using electromagnetic fields, but do not have to be. Also, an understanding of the theorem requires an understanding of the theory of signals and systems and of the concept of signals as vectors in an abstract "signal space." Covering these concepts in a physics class would take it too far afield of the ropes, pulleys, falling bricks, etc. that it must cover to introduce the basic concepts it treats.A "channel" is a mathematical object only: a function that combines a set of inputs to produce an output - not a physical phenomenon at all, but the most information-free abstraction.Abstractions are useful to explain the behavior of physical systems.Many people similarly probably believes that intro physics (college Physics I) uses simple linear differential equations because the universe has been proven to be continuous and differentiable.It is, to a sufficient degree of accuracy for those equations to be useful in everyday life. Despite the fact that freshman physics uses frictionless pulleys, massless ropes, and other accessories you're unlikely to find even in Wile E. Coyote's Acme Catalog, it nonetheless explains a lot about the world, and to a pretty good degree of accuracy. I use those equations frequently when doing carpentry.The other problem with Brett's comments is his statement "from the perspective of the receiver, the undesired signals are simply noise".They are.There is a difference between undesired signals and noise. Noise is technically quite different from undesired signals.Actually, it is not. In fact, as I've already mentioned, the more efficiently a signal uses spectrum, the more it looks like random noise to a third party.What makes noise different is that noise is unpredictable, whereas signals are indeed predictable.I cannot predict when someone else who is "stomping on" my radio signal is going to transmit, nor can I predict what type of modulation scheme he's going to use or for how long he'll transmit. In fact, tailoring my system to reject the type of interference I expect only invites someone to design a different system which does interfere. This has, in fact, happened in the world of unlicensed communications. Some manufacturers have designed systems which intentionally obey nospectrum etiquette, so as to "defeat" systems which try to be courteousto other users of the band or to sidestep interference. The only practical way to design a system, therefore, is to design for the theoretical worst case: Gaussian white noise.Signals are designed to be decoded. Noise is not.One cannot reliably distinguish between the two, because one cannot divine intent. See the comments in my paper at http://www.brettglass.com/ISART regarding SETI and the feasibility of trying to recognize, much less interpret, signals generated by alien civilizations. Also note, again, the spectrum at http://www.brettglass.com/Laramie900.jpg The energy in that spectrum consists entirely of signals which are "meant to be decoded." Yet for all practical purposes it is just Gaussian white noise.If you have two signals plus noise, the limit of the receiver's capacity is not, as Brett would imply in the "undesired signal is simply noise" in S-H theorem: R = W log(1 + S1/(S2+N)).This is not correct, because the second signal is (by your own assertion)not noise.Instead the system of two transmitters and one receiver can achieve a rate of: R = W log(1+ (S1+S2)/N).I have already stated this myself. As I mentioned in my earlier message,techniques which rely upon multiple transmitters (or multiple transmissions) to improve throughput can be accounted for by modifying the signal to noise ratio. Simply adding them together is a best case scenario, though. In most cases, you'll find that the numerator of the fraction is (S1+S2) times some constant which is substantially less than 1. We see this in orthogonal frequency division multiplexing (OFDM), which was originally developed by Paul Baran for use in Telebit's modems and is now used in many wireless systems. But to augment the signal to noise ratio in this way, the transmitters must be transmitting the same information or parts of the same message. This is very different from a situation in which independent transmitters are transmitting unrelated information and are not cooperating.It has recently been demonstrated (using information theory and a conservative physical propagation model) by Tse and others that the communications capacity of a set of radios operating as an adhoc net in a common medium can scale linearly in the number of radios.This model makes a number of flawed assumptions -- most particularly that all of the users of the same spectrum are cooperating. This is not the case in the real world (which brings us back to the politics and policy of spectrum allocation). When it comes to spectrum, I'm not just a theorist and scientist but also an engineer, technician, and user. Today, I was out supervising my crew on two roofs as they installed antennas for wireless Internet. I then helped users at a trailer park connect to a wireless network I set up there. I also do technical support for half a dozen wireless hotspots.(Hotspot support is among the most difficult kind of technical support,because you are usually dealing with a naive user who arrives with a computer system which you have never seen before and which the owner himself does not understand. And because you are usually supporting this person over the phone, you are playing "Blind Man's Buff" because you cannot see his or her screen and he or she does not know the words for the GUI elements which are displayed on it. But I digress... Back to spectrum issues. I see the practical effects of Shannon's Law, the principles of electromagnetic communications,and spectrum policy every day when cordless phones, competitors' signals, and other interference sources threaten the integrity of my network. As an electrical engineer, I'm quite comfortable with the "ivory tower" theory.But it's when one comes down from that tower and climbs a radio tower to build a real world system, one gains real insight into the policies that are needed to best use the valuable resource we call "spectrum." --Brett Glass ------------------------------------- You are subscribed as dpreed () reed com To manage your subscription, go to http://v2.listbox.com/member/?listname=ipArchives at: http://www.interesting-people.org/archives/interesting- people/
------------------------------------- You are subscribed as lists-ip () insecure org To manage your subscription, go to http://v2.listbox.com/member/?listname=ip Archives at: http://www.interesting-people.org/archives/interesting-people/
Current thread:
- more on Response to David Reed and Brett and the FRC David Farber (Sep 08)
- <Possible follow-ups>
- more on Response to David Reed and Brett and the FRC David Farber (Sep 11)
- more on Response to David Reed and Brett and the FRC David Farber (Sep 14)