For finally reporting properly on DIDO. Now please forward this article to every Australian politician.
According to Steve Perlman, his new wireless technology, DIDO, opens up a brave new world of wireless data that promises infinite and endless capacity for everybody. That's got some people excited (so excited that they can do no more than repeat his claims verbatim), while others have suggested that snake oil may be on sale …
Sounds as if you are an anti-National Broadband Network (NBN fibre) punter who's looking for any 'evidence' and anything's OK to feed politicians who wouldn't know facts from crap anyway.
It's 04:30 or thereabouts here and half of my neurons have already closed down for the night but there's enough awake to know that this is (a) huge misrepresentation/oversimplification of a much more technical document, and/or propaganda concocted specifically to mislead/deceive and produce even more FUD in the light of the current NBN controversy, or it's written by a non technical person who has hardly a grasp on the subject.
I'll reread the doc again in the cool light of day just to be sure. However, irrespective of unscientific BS, mumbo-jumbo and political ideology, physics is still physics and it doesn't change its spots for anyone--not even marketing types, spin doctors or politicians.
Facts are facts and here's a few:
1. Shannon's Law applies to any channel no matter how it's disguised, re-badged or what colour it's painted.
2. Repackaging of channels, bundling of them or sending them across different mediums is permitted but Shannon still applies:
3. Restated, this means YOU CAN'T GET SOMETHING FOR NOTHING. IF THE AVAILABLE BANDWIDTH IS INSUFFICIENT FOR A GIVEN DATA STREAM/CHANNEL THEN IT WILL NOT FIT. CODING THEORY SAYS THIS. You either slow the data rate down to suit or you increase the available bandwidth (or, if inefficient, improve the coding/modulation).
4. Modulation systems/schema are not all the same. Just because some encoding schemes are common, it doesn't mean that they are efficient. AM and FM radio use coding schemes that are inefficient, especially so for FM. This means that the same info if coded with a more efficient coding scheme could be fitted into a narrower broadcast channel without losing any info of suffering any loss of signal-to-noise ratio.
5. The same physics applies to Wireless Internet channels as it does to those of radio or TV broadcasting. The difference is that the type of info carried and the efficiencies of the coding (modulation) are different. Newer schemes usually make better use of the available bandwidth but the fundamental rules remain identical. Better encoding allows more info to fit down a channel of a given bandwidth.
6. Newer and better encoding techniques are now possible because of cheap electronics, fast microprocessors etc.
7. In recent years, encoding techniques have been getting better. Nevertheless, Shannon still applies. We are now at a point where squeezing more data through a given bandwidth has just about reached the point of diminishing returns (the Shannon limit).
8. An example of the point of diminishing returns is the 56k dial-up modem. They're now ancient and we've still not seen faster ones of this kind. Simply, getting more info onto an audio telephone line has just about reached its upper limit with this technology (theoretically and practically and so into the future).
9. BEFORE YOU EVEN DREAM OF MENTIONING IT, DSL/ADSL DOES NOT CHANGE THOSE RULES, THEORY AND SHANNON ARE STILL INTACT. 56k modems work WITHIN the audio bandwidth of a standard POTS telephone line whereas DSL/ADSL does not. Like the radio spectrum but to a very much lesser extent, copper telephone lines are capable of carrying channels that use different encoding techniques. The traditional telephone audio consists of a baseband around 300Hz to 3kHz whereas ADSL uses an encoded carrier technology in a similar way a radio station does. Moreover, for various reasons too detailed to mention here, DSL/ADSL is a fudge or kludge, it only works up to about 5km from the exchange whereas the baseband telephone can work 10 perhaps even 100 times that distance. Essentially, the granularity of POTS versus DSL are totally different. In communication terms, we're not comparing apples with apples but with oranges.
10. The same analogies can be applied to wireless. The radio spectrum is capable of carrying any type of encoding--from 10 words-per-minute Morse Code to huge data rates of satellite links etc.--simply, the spectrum is blind to coding techniques.
11. Just because the spectrum is blind to coding techniques it doesn't mean that all spectrum is the same. Higher carrier frequencies are best suited to wide bandwidth channels because there's more spectrum to fit channels within. However, in practice the higher the frequency and the wider the bandwidth the shorter the circuit distance.
12. Existing radio schemas (bandplans) are not perfect and can be improved. Whether DIDO actually does this or not, or whether it's cost effective to do so is unclear as there's no proper engineering data available to us. Mumbo-jumbo will not suffice here, only quantitative specs will do that.
13. The PDF article is extremely sloppy in its nomenclature and terminology. Cellular radio technology (a) reuses the same frequencies repeatedly, (b) where channels are doubled up, wireless devices can slide along to other clear channels, and (c) wireless today uses spread spectrum techniques. Oversimplifying somewhat, this technique allows cell phones to share the same channels but to simultaneously interleave themselves with each other so there's little or no mutual (same/co-channel) interference.
14. There are many ways of packing more channels into a given amount of spectrum but such schemas are not infinitely extensible. Interference, cross-modulation and Shannon's Law guarantee that it's not. In this respect, the PDF article makes outrageous, unjustifiable claims that border on something that a charlatan would write.
15. The ionosphere is a very useful physical phenomenon, but it's a very unpredictable one. Channel (bandwidth) availability is very restricted, path/skip distance cannot be relied upon and fade and noise margins are highly problematic issues. Heavens knows how it would be properly integrated into the proposed schema (we've over 100 years experience of using the ionosphere for communications and these issues are extremely well known).
It seems to me that most of the protagonists in this argument need to get out of the way and let experienced radio engineers set parameters and devise new systems.
An elementary course in radio theory for most rest of you would do much to cut the speculative and unjustifiable claims--most of which seem to be more in keeping with science fiction than having any basis in physical reality.
It doesn't look like complete snake oil - by my reading, it requires one AP per device to get full speed! You can do that without "DIDO" if you just put the devices close to their APs and the APs far enough apart.
The supposed magic here is in computing how the signals from multiple APs will sum at the client points; while there are some big hurdles, it's not inconceivable that someone has engineered a reasonable solution (modulo hyperbole).
Anyone who wants to look in more detail at the "immensely complex mathematics" can probably have a geeze at the patent applications:
Well said. All the posters below should probably slow down a bit and ponder that for a moment.
This is a clear example where Shannon's law does not apply - and like you I suspect it is analogous to what he's seeking to achieve, effectively replacing 'physical geography' with "immensely complex mathematics".
Whether it could work or not, I've no idea - the thing with complex math is it's very easy to miss the little mistakes that nullify the whole thing.
But Joe Regreader can often be as dumb as the rest of humanity - here's a clue folks: Snake Oil vendors do it for a living - I can't think of a single case of someone with a valid, well-earn reputation in a field becoming a snake-oil dealer. It just doesn't happen.
So, I'll wait until the math is properly peer-reviewed - if it holds up, then I am entirely confident that he will be correct in saying that Shannon does not apply, in the specific sense that he meant it.
Wikipedia sez so. http://en.wikipedia.org/wiki/Snake_oil
"Snake oil is a traditional Chinese medicine made from the Chinese Water Snake (Enhydris chinensis), which is used to treat joint pain. However - "
Now I don't know if you're being sarcastic:
"I can't think of a single case of someone with a valid, well-earn reputation in a field becoming a snake-oil dealer. It just doesn't happen."
I can't name a name, but in most science, particularly physics, the brilliant breakthrough is done by a young, smart, sharp brain. But by the time they've done it, they are not so young, smart, and sharp any more. So they're unlikely to be able to do it again. Unless they cheat. It happens.
However, while I'm not qualified to judge the physics and theory in this case, I think the reference to Shannon's law not applying could mean merely that there is good reason to consider that Shannon's law does not limit the efficiency of this device as you would casually expect. One thinks of the admittedly mythical calculation that in terms of aeronautical science, a bumblebee can't fly; but it does, but in a different way. One also thinks of the history of modems as told in Wikipedia
in which the apparently impossible was achieved over and over again, including, again apparently, ignoring Shannon's law to build modems with a speed above 33.6 kbps - and wasn't that done too by using superior data processing power in the data centre?
On the other hand, that correspondence also makes sense if the present story is deliberate hokum but intentionally based on the history of modems.
If the capacity is theoretically infinite, that just means that another constraint becomes applicable, such as electrical or radio interference from the microprocessors themselves, passing car engines, or the Sun (or the Big Bang).
> Because of this, each wireless user can use “the full data rate of shared spectrum simultaneously with all other users”, and therefore, “Shannon’s law does not apply”.
Either the spectrum isn't literally shared, or Shannon-Hartley theorem does indeed apply. There are a whole pile of tricks you can use to avoid sharing spectrum: create small cells that don't overlap, use highly directional antenna's, use phase array antenna's to create a virtual directional antenna, bump up the signal strength to raise the signal to noise ration, figure out a way to reduce the noise by say running your signal through very thin glass wave guides. If Steve Perlman hasn't invented a new one of these it is snake oil. Given the way the paper is written, I'm betting on snake oil.
Notice that Shannon-Hartley is called a theorem, not a law. That is because isn't one of physics model's validated through repeated observation. It is actually something stronger - a mathematical fact. Saying it is wrong is like saying 1+1=3.
...stab at explaining what's happening.
In a beam forming radio, two or more antennas are used. Traditionally, these antennas are located at precise spacing from each other in order to simplify the math needed to send the correct waveform to each antenna. However, the antennas may be located at random distances, it just requires more complicated processing of the waveform.
DIDO centralizes the calculation of waveforms for each antenna, allowing antennas to be widely and randomly spaced, yet form a "virtual beam" at the receiver.
Please look at:
for a look at the theory of beam forming. Then at:
for actual, if unstructured, performance tests.
Finally, let me repeat that photons do not interact with one another. Additive and subtractive effects take place only in the electrical currents induced in the receiving antenna. DIDO simply (actually not simple to do, bur simple to understand) calculates the desired resultant waveform at each antenna.
I recall around 2003 seeing a panel array of 128 slot antennas that was capable of 200 simultaneous WI-FI connections.
It boggles the mind, but beer is the great unboggler.
Zeke, you nailed it. It seems to be a beamforming antenna, using centralized processing rather than a processor next to several nearby antennas. So it's combining Distributed Antenna Systems with MIMO beamforming.
There's no Shannon violation, because Shannon's Law refers to actual gaussian noise, unpredicatable stuff. In DIDO, the assumption is that everything is correlated via one CPU, so the unwanted siganls are known, not true noise, and thus can be mathematically cancelled. Of course this assumes that the properties of the path are well known and constant. Don't try this with mobility.
So the claims are grossly overblown and the actual ideas are old, just repackaged. It could theoretically work a little, but in practice won't be all so special.
Anything claiming "infinite" anything or claims to break known laws of information are clearly bogus and are a scam.
(*) That guy didn't end up in court because he didn't understand CS but because he was deliberately defrauding investors.
I think that for the next few years I'll be taking the physical connection and utilising any wireless parts within the home if necessary. I haven't seen anything that would make me want to go all wireless especially if the latency and network capacity on my mobile phone are anything to go by. New wireless technologies seem to come and go quite frequently.
So, by providing one wireless station per user, each user gets the full wireless station bandwidth? Amazing!
Rather pointless doing the maths to arrange that interacting AP signals result in the user getting a clean signal though. If the range needs to be short enough that the number of APs in range isn't beyond any calculation not requiring the rest of time, and you have 2 bandwidth hogs sitting at the same cafe table, maybe we could do something revolutionary like provide a physical cable or fibre access point for them.
I well remember the calculus exam question from my City and Guilds Telecom's Principles exam - 'Derive from first principles the instantaneous value of an FM wave', this would be a nightmare!
The thing is, something HAS to look like making his ridiculous bandwidth hogging OnLive scheme actually achievable, I'm smelling a scam before he sells his shares.
I should also point out that this scheme is all one way AP(s) to User, unless you are packing a quantum hyper-computer and multiple directional transmitters, you can't work this trick backwards and reassemble the signal back at the data centre.
Probably not a problem for many bandwidth-assymetric uses though.
Of course if each user can get the exclusive attention of one AP then you might get high TX bandwidth, but the point is supposed to be that when in contact with multiple APs, 'magic happens'.
Giving the benefit of the doubt to this ad-lib paper, multiplexing (even "distributed") does not invalidate Shannon's Law.
The fact that this paper claims so makes me question the credibility of the rest of it. I think this is more of a vision than a technology. It's quite easy to say "low latency" and pepper a few buzz words (cloud anyone), but there are some basic problems any real system has to deal with that are simply not even vaguely addressed, as if they don't exist. A gloried explanation of channel estimation does not a new technology make.
So far DIDO is by the marketing department, for the VCs.
"5. The mathematics behind DIDO is too long and complex for mere mortals to understand."
Yep, nothing like a reputable scientist refusing to put a theory or plan up for review by their peers.
One can only presume it's because it doesn't actually exist, or it's voodoo mathamatics.
There are patents to protect work like that. Even if the patent system is broken.
"which mediates all communications between user and content provider." ..... Mediates suggests an element of third party or DIDO vetting of capability. Would facilitates be an available option and a special intelligence service arrangement providing for secret stealth communication between thin client users/leading vertical systems suppliers and content provider/base source code kernels? Or would that be a question too far and best left unasked, for quite obvious reasons.
Indeed, one could even very reasonably expect a positive negative response to suggest the option is not provided even though it is probably definitely maybe available in future developments of contiguous IP ..... and an ASPen snake-oil for lubricating the wheels of reality in Live Operational Virtual Environments?
That is surely Intellectual Property you are to die for, because it is so good at providing invisible, intangible, invincible presence which is virtually real in InterNetworking Global Operating Devices.
"The idea that DIDO instantly makes all fibre networks (including the NBN) obsolete overlooks a host of issues that aren’t answered in the Perlman white paper." ...... Given the fact that SMARTer IT use can render any and all networks obsolete over any issue, sensible failsafe security considerations would explain and be the reason for the "a host of issues that aren’t answered in the Perlman white paper" situation, in what would most surely, because of what can be claimed to be easily done, most definitely be highly classified way beyond any mundane or urbane pedestrian/terrestrial Top Secret/SCI Need to Know.
"As for DIDO itself, if Steve Perlman wishes to convince the world that it’s a genuine breakthrough, a lot more detail is needed." ..... That would be a something of an idiotic vanity project, would it not, for the world is full of folk with very little interest in anything of great note, and as a consequence are easily led to believe whatever frivolity may be presented before them? It is a systemic weakness which can be ruthlessly exploited most sublimely.
"2. Because of this, each wireless user can use “the full data rate of shared spectrum simultaneously with all other users”, and therefore, “Shannon’s law does not apply”."
Talk about marketard-babble! This one has me horse-laughing ... and wondering if I should get in on the ground-floor, to separate the fools from their money.
Sadly, I'm not a con-artist, so it probably won't happen ...
Still ::giggling:: at the stupidity, though :-)
Well that's perhaps a bit exaggerated. However with n receivers and n transmitters you can have n transmissions independent of each other. It's called space multiplex.
Or if you want to look at it in a different way. The channel capacity is given by the signal to noise ratio. Noise is, on a good system, only thermal noise which is only dependent on the temperature (and bandwidth). Now if you have n antennas, the signal rises with n^2 since it's correlated while the noise power only rises by a factor of n. So your SNR gets better by a factor of n, therefore you can transmit more data.
The days of pure thermal noise have long gone. These days most of the RF noise is man made.
The concept of n transmitters and n receivers would work ok if you had n channels. However if all the receivers and all the transmitters operated in the same channel then what is signal from one transmitter is noise from another.
So you can beam form with n transmitters to one receiver so that receiver will get a perfect in phase maximum power signal and all the other receivers will get out of phase overlapping signals that will appear to be noise. However you will not be able to simultaneously do this from all transmitters to all receivers in the same frequency space. Just think 100 transmitters and 100 receivers, each transmitter will need to send out 1/100 of its signal adjusted in phase for each receiver which means each receiver will receive 100 signals from 100 transmitters with 1/100 of each signal in phase for that receiver and 99/100 of the signal out of phase, adding them all together gives you 1% signal and 99% noise.
So, for this to work it will be time multiplexed and the total channel capacity is actually max capacity / number of active receivers.
Essentially it's something which became obvious with the first "modern" analog cellphone systems in the 1970s. There you had multiple receivers to receive the signal from the mobile station, the network would then always use the strongest signal. This is also called diversity reception, in this case even macro-diversity.
From that point on you can go a step further and combine the signal of multiple antennas by means of beam forming. This is also known for decades and in fact when we discussed UMTS at university, the professor told us that those base stations were already in development. The next obvious step is to simplify the base stations to a point where you only have a simple transceiver and do all the beam forming and data processing centrally. This is something the UMTS network infrastructure was made to be turned into.
So it's obvious. Yes it works, but it obviously cannot do miracles. It's a sane idea to do, and I'm sure it would have been implemented in UMTS independently of that patent troll here.
You can obviously use the estimations of the impulse response of your channels to do a quite good estimation of the position of the transmitter.
This can also be used to provide position dependent services for the user of that transmitter.
And if you know the position of the transmitter you can also optimize the downlink for that position and improve throughput there, too. For example by delaying or de-convolulating the individual signals to cause constructive or destructive interference at certain points in an n-dimensional space.
There, just destroyed a few follow up patents. We really need an archive for obvious ideas to prevent obvious patents.
The Shannon-Hartley theorem (quoting freely from Wikipedia) "establishes a bound on the maximum amount of error-free digital data (that is, information) that can be transmitted with a specified bandwidth in the presence of ... noise interference, assuming that the signal power is bounded, and that the Gaussian noise process is characterized by a known power or power spectral density."
In the case of DIDO, much of the 'noise' experienced by one user of the channel is not random (unpredictable) gaussian noise, but is actually the result of the combined signals from other users. Since DIDO knows (almost) exactly what these signals are, it is reasonable to expect that it could improve on the information limit specified by the theorem, although it is hard to know to what extent. The Wikipedia article suggests as much: "If the receiver has some information about the random process that generates the noise, one can in principle recover the information in the original signal by considering all possible states of the noise process". In practice, the limits are likely to be determined by the amount of true (unpredictable) noise in the channel, the sensitivity of the receivers, and the fidelity with which the transmitters can reproduce the ideal signals specified by DIDO. Tests with just 10 users are not going to establish the real limitations.
Exploting the relative independence of different signal paths between users and access points is also sensible, but there does not seem any reason to expect it to be massively better than the multi-path technologies already in use.
Combining signals from multiple sources to achieve the desired signal at a point is far from new - anti-noise technology has been doing it with sound waves for many years. However, and this is the oiliest point, to achieve this some of the peaks of the transmitted signals will need to be higher than the peaks in the resultant signals (this is obvious if you think about the reverse operation - if two users transmit their desired signals, some of the peaks at a receiver will interfere constructively). Thus to stay within regulatory power limits, the effective power experienced by a user in the channel must be reduced, thus reducing the channel's information-carrying capacity and hence reducing some (all?) of the gains claimed for the system.
On top of all this, DIDO requires a complete lock-in on all users of a given channel in a given area. Unless it can demonstrate huge performance benefits, such a degree of control is unlikely to be acceptable (although perhaps more likely in certain dedicated application areas, as the DIDO brief seems to suggest).
Wireless can't replace Fibre except in either Broadcast or very sparse Rural applications (5 users).
Actually the Shannon Boundary and Shannon Limit derived from Shannon-Hartley Theorem probably are "laws". The Shannon-Hartley Theorem was a theory, but it's probably a law.
AirPoints don't replace ISP fibre, DSL or Cable connections.
My understanding was that the data doesn't all go back to the data centre( That would be a massive and stupid bottle neck) but the analysis of the link is done at the data centre (why?) and then the parameters to use are sent to the AP. Which also sounds pretty stupid. Why not an SoC?
It's fake snake oil. I've been working with wireless systems a long time and even all the "real" systems are heavily lubricated by snake oil. E.g. 42Mbps 3G/HSPA+ when the capacity *IS* shared among all users and at typical distance and signal you'll get about 4Mbps in a 5MHz channel if there is no-one to shared it with.
The major issue with Mobile outdoor bases or indoor AP to laptop/gadgets turns out not to be nuances of modulation and coding. But aerials. If your house has a directional aerial on the chimney for a fixed wireless system, if everything else is equal, that will have x8 capacity in same spectrum (Shannon applied carefully) compare with ground level indoor mostly omni-directional; aerials in a laptop or gadget. No one wants a giant motor driven aerial on their mobile phone or laptop, so such systems with mobility are always inferior to fixed outdoor directional aerial systems.
Some more articles here : http://www.techtir.ie/comms
It seems that I have to send all the data to a server to be be processed (presumably via my exceptionally fast fiber connection) before I can send it to the client devices over the air waggle thingies.....
So to support 10 devices chatting with my local NAS at >52MB/s I have to have a gigabit net connection? That seems freaking insane.
Oh, and the other question is, how do all those remote devices talk back to me without treading on each other? It seems like we'd end up with a seriously asymmetric connection, possibly to the point where, if there are lots of these beasties in the environment I might never actually get a response from my mobile device... unless they also had nice gigabit connection back to the DIDO server farm.... oh what?
In order to overcome mathematical limitations we will now throw huge computational resources to produce the equivalent of a dynamic 3D hologram so everyone has their own view of the picture.
It only takes lots of radio sources controlled by us. If you wish to partake you will need a fast connexion to get back to us! Obviously you do not have our computing power or algorithms.
Where I come from this is called broadcasting. Bert
If the signals are separate signals (beam forming etc) then each signal is subject to Shannon Boundary.
If the signals share the same beam, then Shannon STILL applies. The total information bandwidth is limited. The bandwidth can be shared by time, frequency or code. (or any combination). Those are multiplexing techniques. As CDMA has shown, code division multiplexing is poorer than Time or frequency. Absolutely perfect signal recovery (i.e. better than Rake) only makes it as good as other schemes. TDMA wastes space on guard times between channel. FDMA wastes lots of space on frequency guardbands. Datamultiplexing (as in DVB) is 100% efficient, but only works for downlink, up link has to be CDMA, TDMA, FDMA or OFDMA. The latest technique is OFDMA as the guard timing is very low if the stations are stationary or if there is no multipath. You might use 3,500 carriers at slower symbol rates and for uplink vary the number of carriers assigned to each station. Downlink uses all carriers and Data-multiplexing.
You can't get something for nothing. The DIDO sounds like the wireless equivalent of "over unity" machines. No doubt it can work, but no more than 1% or 2% better than the best OFDMA systems with beam forming. If it's 5% better overall than LTE, Wimax or Flash-OFDMA I'd be amazed. If it seems 10% better then its a scam and there are hidden wires.
"The big issue for lots of internet users is the idea that if they’re going to use a wireless connection, they will have to hand all of their communications off to one organisation for pre-processing."
Well, it doesn't require that at all, really. The way I read it, the DIDO base stations would be, say, per-apartment-building or per-block; basically, they cover the area where currently your wireless signal overlaps with 40 other wireless signals. They seem potentially to be able to be quite small scale. So there could be lots of different organizations providing DIDO base station service.
The privacy angle also seemed an obvious one at first glance until I thought "But...we already trust one organization with implicit access to all the data we transmit over the internet" - our freaking internet service provider. I can't see any difference between an organization that provides DIDO service and an ISP, in practical terms. I think the author's Glorious Future Vision is that they would become one and the same thing. Hence, no increased privacy issue.
The Magical Mathematical Hand-Wave remains the biggest issue: as described the system sounds great, but they already admitted they've tested it with like ten devices max, and it isn't terribly reassuring that they don't want to disclose the actual mathematical basis of their claimed ability to beam some sort of super-signal which looks to hundreds of different APs like the exact signal they need to see...