Harnessing Interference For Faster Wireless Data

holy_calamity writes "Inventor of the Quicktime codec Steve Perlman has unveiled a new wireless technology he claims can deliver thousands of times more bandwidth to mobile devices than existing technology. Each user is served by multiple transmitters, which send out waves carefully designed to combine into a data signal only at a device's location. That technique enables every user to be targeted with a signal with the same total bandwidth that would usually be shared between users, says Perlman."

Comments

Can still charge

[Toe, The]

If companies like AT&T and Verizon come up with a good way to provide boundless bandwidth... what do you think are the chances that they'll stop charging for high usage? 0.005% maybe?

Re:Can still charge

[ScentCone]

Right, because the back end of their system, and all of their peering connections to all of the other internetworked networks comes at no charge, and is infinite.

Re:Can still charge

[rojomojobojo]

Yeah, in the beginning. Once they have their network up, pretty much everything is pure profit.

Re:Can still charge

[Anonymous Coward]

those are fixed costs. Paying $20 for 2GB, but $120 for 4GB isn't fair. They don't charge you less when you use only 150MB vs. using 750MB.

Re:Can still charge

[Anonymous Coward]

Yeah, because the network magically runs itself. And they don't have any employees, taxes, rent, bills, upgrades, maintenance, marketing, etc.

Re:Can still charge

[Lunix Nutcase]

Right, it's not as if they inflate their actual costs or anything. I mean it's makes perfectly logical sense that a 200MB data plan from AT&T costs $15 while 2GB costs $25. Care to explain to me why the former plan costs nearly 7x more per MB than the latter?

Re:Can still charge

[Lunix Nutcase]

No, but the fixed costs of the infrastructure eventually ends up being far exceeded by the revenue coming in which is his point. Or do you care to point out how 1GB of "overage" data somehow costs AT&T 10 times more than a "regular" GB of data.

Re:Can still charge

[Anonymous Coward]

Nope, still not buying it. Note how the GP used "upgrades" in that list of things costing money. Given no amount of bandwidth is truly "boundless" or "infinite", some joker will come up with some way to saturate the line 24/7 and complain that they neeeeeeeeeeeeeeed to keep it up, complete with all whining about "my rights", ambiguous contract definitions they assume are in their favor, etc, etc, etc, until AT&T needs to upgrade the connection.

I assume that, between the introduction of the newer whiz-bang faster connection and the point at which certain people decide they need to saturate the line, this will take around a day, maybe two. So out come the upgrades to the system (costs money) for another day of use before the entire cycle starts again (costs money each cycle), until they reach the limits of current technology and need to research something better (costs more money) to begin the cycle anew (costs increasing money ad infinitum)...

Re:Can still charge

[GooberToo]

So you're anti-capitalistic now? You're socialist?

Why is it so many /. posts demand socialistic freebees and but a capitalist paycheck. Its should only be free (and socialist) when its out of someone else's pocket.

I guess the entitled generation has spoken.

Re:Can still charge

[Anonymous Coward]

Yeah and if you are Go-phone prepaid customer forget it. the most you can get is 200MB of data and its the cost I pay for my 2GB currently 25 dollars. So they screw the people who are down on their luck. They should have the same data plan for the same price.

Re:Can still charge

[ScentCone]

Care to explain to me why the former plan costs nearly 7x more per MB than the latter?

Because you're making the common mistake of not understanding (or pretending not to) that both plans carry administrative costs and overhead that costs a lot more than the bandwidth. You're thinking that the only thing built into those prices is the actual bandwidth. Which is exactly wrong.

Re:Can still charge

[sjames]

Because we have to eat until the day comes when we get the socialist freebies? Because it's necessary to counterbalance all the corporations succeeding at socializing costs while privitizing profits?

Because the 'invisible hand of the market' is supposed to drive the retail price down to the marghinal cost of production, but it never seems to do so?

Because we figure that with all the subsidies the taxpayers have given the telcos we should own them by now?

Take your pick.

Re:Can still charge

[chill]

No, he was assuming that the administrative costs and overhead of the two data plans are identical. Considering the only thing that changes is the bandwidth or data cap, it is a reasonable assumption. Actual operation of the service is automated and doesn't have proportionate cost increases.

For example, a sysadmin that can manage a 100 Mbps Fast Ethernet switch doesn't cost 10x less than one that can manage a GbE switch. Nor is there 10x the work involved.

Yes, there is a different capital cost, but ongoing administrative costs are the same.

Re:Can still charge

[SonicBurst]

Let's agree that each plan has equal administrative overhead costs, and that for each plan this cost is $10 (actual admin cost should be nowhere near that, but hey, this AT&T we're talking about). So, that leaves $5 for 200MB of data, and $15 for 2GB of data. So, 1/3 the cost for only 1/10 the data? OP's point still stands. They're screwing you on the high(er) end plans, but they're screwing you even worse on the low end.

Re:Can still charge

[Anonymous Coward]

So you're anti-capitalistic now? You're socialist?

Why is it so many /. posts demand socialistic freebees and but a capitalist paycheck. Its should only be free (and socialist) when its out of someone else's pocket.

I guess the entitled generation has spoken.

You might want to check into what titans of capitalist economic theory (like, oh, say, Adam Smith) have to say about how pure free markets actually work in practice. Here's a hint: Smith would think all these modern "libertarians" are a pack of fools. He thought regulation was very much necessary, because most markets aren't as free as they ought to be. (As in: most markets cannot function with the highest economic efficiency without some form of regulation. Where by "efficiency" we mean that the behavior of the market should roughly optimize economic output and distribution for the greatest good of society.)

Telcos (including wireless telcos) are naturally monopolistic businesses (very high barriers to entry, easy for incumbents to block new entrants, inherently lower costs per customer the larger the business gets). This is precisely the recipe for a market which needs heavy regulation -- without it, natural monopolies will screw society every time. The famous "invisible hand" is not an infallible mechanism which works equally well under all circumstances; sometimes it needs guidance.

Which is why you get stupid stuff like AT&T charging 10x more for 1GB of overage, and, due to a political climate in this country where people like you brand anyone to the left of Randroidism as evil commies, often nothing much gets done about it.

Re:Can still charge

[ultranova]

Why is it so many /. posts demand socialistic freebees and but a capitalist paycheck.

Because capitalism assumes low barriers of entry to a market (so that if AT&T charges too much, new startups pop up to eat their market share), while phone industry has extremely high barriers. Because the assumptions behind capitalism don't hold for phone industry, it would be stupid to insist running said industry by it, unless one has some kind of ideological commitment to capitalism - and we only need to look at the ruins of the USSR to see what happens when you insist that ideology should take precedence over reality in managing your economy.

Oh well. Maybe the lesson finally sinks in once the US collapses too. But more likely, capitalists will blame socialism and socialists capitalism when in reality it's the blind adherence to One True Economic Theory and trying to force it on everything, whether it makes sense or not, that's the problem. Well, that and plenty of rich people ruthlessly exploiting that blind spot to get richer at other people's expense.

Its should only be free (and socialist) when its out of someone else's pocket.

Public utilities are so called because they can't operate without public support, such as right-of-way. Because they require public support, the public also gets a say in how they operate and at what price. Which is only fair, unless you are suggesting that these companies should get freebies from everyone else's pocket?

I guess the entitled generation has spoken.

Let's hope it speaks more often, then, so that we can start weaning people off the strange notion that society exists to help corporations make profit, rather than the other way round.

Re:Can still charge

[Anthony Mouse]

Given no amount of bandwidth is truly "boundless" or "infinite", some joker will come up with some way to saturate the line 24/7

Nope, not buying it. The demand for bandwidth is not infinite. Just look at ethernet in a typical business network with a normal file server, traffic to the internet, etc. There is just no scarcity there anymore: Even if you have a "fast" file server, the bottleneck is the disk array, not the 10GbE uplink to the switch. Nothing normal people do even remotely taxes a network with gigabit to the users and 10GbE to the servers.

The fact is that consumption does not increase linearly with capacity because the more capacity the network has, the more likely it is that the bottleneck lies somewhere else, and increasing capacity only causes network utilization to increase where the network was previously the bottleneck. There comes a point where increasing capacity will not cause any increase in utilization whatsoever by 99.9999+% of users. Even if there exists a couple of asshats who just consume all available bandwidth for sport, there are very few of them and their cost can be amortized into pennies across the millions of normal users who don't do that.

Re:Can still charge

[shaitand]

AT&T is nowhere near saturating their lines, nor are any of the other major carriers. They cried wolf in Canada and Canada made them actually reveal the data and nowhere in their network were they anywhere near 50% saturation at peak.

Re:Can still charge

[Bitmanhome]

You fail economics. The "chance" is zero, because it's not random. AT&T strives to make as much money as possible, unless competition forces them to do otherwise.

Re:Can still charge

[Anonymous Coward]

The chances are zero. Rational, profit-maximizing firms will not pass on a revenue opportunity. It's basic economics.

Re:Can still charge

[xelah]

The economic cost is different in the short term and long term. The financial cost to the company will be different to that because markets and technology never work properly, but it's probably a good start.

In the short term the economic cost to using extra bandwidth is zero until capacity is reached. So is the financial cost to the company. The infrastructure and organizations have to be maintained whether is being used or not. When it reaches capacity the economic cost is the cost of failing to deliver the data which now doesn't get through (plus the cost of slowdowns and any other inefficiencies which come about because of the way TCP/IP works). If the system is economically efficient the person who values the transmission of their data the most has their data go through, and those who value it less than some threshold don't. TCP/IP and this market aren't efficient like that so the cost will be more. The short term financial cost to the company can come only from other people not using their overage because they're the only people who'd be paying them extra. Because there will inevitably be a mix of overage and non-overage data not being sent you will inevitably be paying more for overage than the average short term financial cost for as long as they're charging you a fixed amount at all. Or, to put it another way, the financial cost to selling that bandwidth to you is the cost of not being able to sell it to someone else.

In the long term the economic cost of using more bandwidth depends on when you use it (is the link at capacity at that time?), how much others who'd like to use the bandwidth instead of you would be willing to pay, how much people who wish to use bandwidth at other at-capacity periods but can't would be prepared to pay and how much it costs to upgrade the infrastructure. This cost will always be less than the short term cost - if the short term cost would be sufficiently high then, given sane financial incentives, the network would be upgraded because that's cheaper. Financially, there will be times when you're costing them nothing (you're only using it at not-at-capacity times). The rest of the time you're causing them either lost overage from others and lost subscriptions, or you're causing infrastructure costs, depending on which is most financially advantageous. If there's only a short period of hugely over capacity demand and you're part of it then that may cost them more than if there's a long period of a little over capacity demand. In the first case c more capacity gives c * (short time) more bandwidth they can sell, in the second it gives them c * (long time) more bandwidth. But the cost of lost subscriptions may be different, and we don't know what that is.

They can't charge you what it costs because what it costs is insanely complicated and depends on what long term commitments to buy bandwidth you're prepared to enter in to. Even they probably don't know what it really is. Either way, you're most certainly not going to work it out in a 'yes it is' 'no it isn't' argument.

Re:Can still charge

[ConfusedVorlon]

of course - utilisation might well vary according to plan.
and cost is really only driven by utilization during peak periods.
more importantly, this cost has to cover the cost of customer acquisition and infrastructure build.

here's the reality though;

Companies charge the fee which they think will maximize their profit.

They don't really know what that is, so they wave their hands a little (actually a lot). They try to charge fees which will cover their investment and generate a profit on each segment whilst building market share. So, people who use more bandwidth get charged an amount based on
-a guess of what most/many of them are willing to pay
-some comparison to what other companies are charging
-how excited the some middle manager is about growing market share this quarter

Re:Can still charge

[sonicmerlin]

I remember hearing about this. Do you have a source by any chance?

Re:Can still charge

[sonicmerlin]

Uh... they work out the costs: http://www.dslprime.com/dslprime/42-d/4148-costs-and-caps

Reliablity?

[rossdee]

Doesn't sound very reliable to me - and what if you move slightly?

Re:Reliablity?

[Anonymous Coward]

Something tells me it doesn't need to be that precise. Sending waves in a general direction is good enough.

Re:Reliablity?

[tiberus]

That depends on how fine the resolution is and how you define 'slightly'. I'd guess (in a totally unscientific manner) your sweet spot would be at least a 1-2 meter diameter sphere. The problem becomes more interesting if you are a moving target.

Re:Reliablity?

[Anonymous Coward]

Wow good job, you totally refuted people who know what they are doing. No way people who spend years on the subject would have thought about that before. Guess they should have talked to you earlier.

Re:Reliablity?

[Mia'cova]

The article says their implementation should be sensitive enough to deal with movement, even driving. Not sure if that claim will hold up but at least they're thinking about it.

Re:Reliablity?

[Anonymous Coward]

I bet it won't stand up to dynamic multipath reflections. Not much does.
I had a large construction crane about 0.5Km from my house that was about 30 degrees away from the backside of my TV antenna. At night the prevailing winds weathervaned the crane back and forth so it reflected the local broadcasts now and then. Whenever the crane moved, reception dropped out until it stopped moving and the dynamic equalization retrained.

If you live in a house by a road with cars and trucks driving by you will get massive dynamic multipath changes at all the licensed data frequencies.
This will kill his plan dead unless there is a constant stream of packets from the user's antenna with the data coming back in short bursts trained from the just received transmission.

Re:Reliablity?

[Anonymous Coward]

In his June presentation, Steve Perlman said these DIDO radios are VERY simple. In the whitepaper, they say a data center handles the APs and modulating the wave forms to account for interference from other DIDO APs. I'm pretty sure the data center monitors the signal for changes in the signal strength of nearby APs and changes the modulation on the fly.

At least, that's how I'd do it if I were an engineer of any sort.

Re:Reliablity?

[Brian the Bold]

As an RF engineer, when I hear people mention the words simple and radio in the same sentence I smile inwardly and anticipate a project that gets to the desperation phase more rapidly than usual without any design input to allow it for tuning the performance of each circuit block.

In short, radio is never as simple as you think it is.

Re:Reliablity?

[Anonymous Coward]

The system can calculate gradients and invert matrices faster than you can run. Even your low-power 3G handset continuously does the same while trying to cope with varying channel conditions.

Re:Reliablity?

[Penguinshit]

In the 90s, Arraycomm had developed a technology that relied on multipath and was capable of putting a spherical signal in less than a half-meter of space. This technology, called "Intellicell", was used first to nearly quadruple the capacity of Japan's PHS system (tiny phones meant to be low-mobility additions to wired phones, serviced my microcell stations located on metropolitan buildings). This was later expanded to a metropolitan data service using more typical cell stations promising over a megabit/sec to each user (in 1999). Single test stations around 2 linear miles from the 2-story tower in an urban environment were able to pull over that rate. US carriers were not interested in rolling out this data service so it was sold to a company in Australia.

Re:Reliablity?

[Anonymous Coward]

I agree, in principal. I definitely don't have the credentials or knowledge you do. I was simply putting forth the proposition that instead of including software and hardware in the access point for error correction, channel sharing, encryption and such, they do all these tasks in software at a central location. I'd compare this to a mainframe and dumb terminal model. The radio itself is simple and only transmits exactly what it is told to by the data center. It passes along all reception to the data center without processing, it seems. The data center that drives the radio is not conducting simple calculations by any measure. The concept reminds me a bit of software defined radios in some ways.

Does this conjecture sound agreeable?

Re:Reliablity?

[Anonymous Coward]

Given that in normal MIMO setting you need to take account other objects moving, not just yourself. Chop down trees (or disable wind) and prohibit trucks on streets.

Security?

[WPIDalamar]

Does this have implications for enhanced wireless security? A wireless signal that can only be received in a specific location seems like a valuable thing.

Re: Move Slightly!

[TaoPhoenix]

"I'm sorry, I thought I was downloading a song by a CC licensed garage band, but I moved slightly and wound up with Britney Spears. So Sorry!"

New ? Hardly.

[billcopc]

802.11n already does this, they call it "beam-forming". Cisco features it in their high-end access points, using multiple antennae to send the same payload but with varying phase shift, which recombine at the receiver to produce a stronger coherent wave.

I love how the summary introduces him as the "inventor of the Quicktime codec". Yeah, he provided the RPZA ("road pizza") codec, which is so damn simple it made Bink Video look like fine art, back in the day.

Here is a more technical breakdown:

[EvilJohn]

http://www.eedailynews.com/2011/08/wireless-inventors-hype-dido-no-not.html

Re:Here is a more technical breakdown:

[MikeDemler]

Thanks for sharing a link to my story on Steve Perlman's DIDO hype, on the EE Daily News. Since writing my report I have spoken with several communications experts regarding this "invention". In a nutshell, the general consensus is "no technical merit".

Interference from other sources is a killer

[CycleMan]

Not mentioned in this article, but called out elsewhere in the press surrounding this, is that this new interference formula only works where nobody else is broadcasting. This can't be used in the existing wi-fi spectrum, for example, because the interference from non-DIDO devices will corrupt his receivers. Unlike FM, for example, which grabs the strongest local signal, this tries to grab and combine all signals under the belief that they will combine properly. If anyone else is emitting on that same spectrum (intentionally or not), it will be troublesome.

Re:Interference from other sources is a killer

[uid7306m]

Nope. Sorry. You're wrong. Electromagnetic waves add very nicely, so that your signal remains there, even if many other signals are simultaneously being transmitted.

The overall idea is fine, in principle. As other people have said, it is 802.11n beam-forming on steroids. If you had 1000 transmitters, and if you could know the exact time delay and attenuation from each of those transmitters to your cell phone, then (indeed) you could make them all add together precisely where your cell phone is. Elsewhere, they would (on average, mostly) cancel out. As advertised, you would (indeed) get a substantial improvement.

But not a *huge* improvement. With N=1000 transmitters, the signal at your cell phone would be (e.g.) 1000. Elsewhere, the signal would be (in the same units) typically sqrt(N) or about 30. That value of 30 is the noise level for eveyone else, and their transmissions give a similar noise level to you. So, 1000 transmitters gives you a signal-to-noise ratio of 30 times better than one transmitter.[*]

[* ...actially, if you're really gonzo, you can adjust all the transmitters to make their signals cancel out exactly, at everyone else's cell phones, so long as you have more transmitters than cell phones. In principle. But I don't think anyone is seriously proposing that...]

Now, you plug that into Shannon's equation, where the data rate is B*log((S+N)/S), B is the bandwidth, and S is the signal and N is the noise. If you start with S=N=1 which is not too far off for a normal system, and change it to S=1000, N=30, you find that the data rate increases by a factor of 5. Which is good and impressive.

But the problem is that it is very hard to know the attenuation and time delay precisely.

So, anyway, it's not a bad idea, but it's hardly new. It's also not easy to implement, and even if you implement it, you won't get factors of 1000 improvement.

Re:Interference from other sources is a killer

[CycleMan]

Based on what I read in other news articles, Perlman himself says that interference from others is a problem.

'Perlman estimates that the first commercial use of DIDO technology could come as early as the end of next year. But even then, the first deployments are likely to be outside of the United States.
In a DIDO system, a data center on the Internet determines the wireless signals that each transmitter will send based in part on the location and number of other DIDO transmitters in the area. In order for the data center to know what the resulting interference patterns will be, there can't be any other sources of signals outside those generated by the DIDO transmitters.
What that means is that a DIDO system would have to be used on currently unused or completely reclaimed spectrum. So DIDO won't be improving your Wi-Fi or 3G experience, because those parts of the spectrum are already crowded with transmitters. It's more likely to be embraced in the near term in countries that have a lot more unused spectrum than does the United States, Perlman said.' (Source: San Jose Mercury News, 8/3/2011 http://www.mercurynews.com/business/ci_18603178)

So Perlman himself says that, yes, in a closed system, his design works -- but only in that closed system. Given that some stray RF emissions are generated from various things (as we know from the amateur radio folks whenever BPL comes up, for example), there is an interference problem.

[* ...actially, if you're really gonzo, you can adjust all the transmitters to make their signals cancel out exactly, at everyone else's cell phones, so long as you have more transmitters than cell phones. In principle. But I don't think anyone is seriously proposing that...]

From the above-mentioned article, Perlman says you do not need more transmitters than cell phones. He suggests a 1:1 ratio instead. "To offer everyone on a DIDO system the maximum data rate and the best experience, a DIDO system would require one transmitter for every user on the system."

Re:Interference from other sources is a killer

[Yamioni]

What that means is that a DIDO system would have to be used on currently unused or completely reclaimed spectrum.

Sounds to me like this would make it trivial to design cell blockers for say, Theatres, Libraries, Museums, etc. Or even for someone to make an even lower powered one to shut up that asshole sitting behind you in a restaurant blabbing at 80db who just won't shut the fuck up. Obviously they'd be illegal for personal use, but damn they'd be satisfying to use at times...

Nice idea, now for the real world

[Anonymous Coward]

It's a very strong novel idea....it has some huge practical issues to solve before it can be considered a possible basis for products. Off the top of my head just a few:

It requires ALL the transmitters in a piece of spectrum to cooperate together...there is no coexistence of multiple independent DIDO systems within the same spectrum without classic Shannon law rules. not too many chucks of spectrum work like that right now....

Implicitly all transmitters in the downstream (Think base station to consumer) direction implicitly have access to all the raw data to be transmitted or a premodulated/encoded version of the same. This implies a back channel. Now consider the upstream. Same problem, every handset transmits everyone's data...where's that back channel?

SGP's claims of incredibly low latency would have to stem from the removal of the temporal diversity that's present in current wireless standards (Forward Error Correction)...i.e you have to receive a relatively large chunk of encoded data over time before you can extract an error corrected payload from it.....DIDO relies on "pre-correcting" for "errors" in the received signal through a characterization of the channel that's periodically refreshed. However sources of dynamic interference from non DIDO sources can still interfere with reception of the signal within the periodic re-characterization. Can FEC really be discarded in the real world?

That will do for now....

Re:Nice idea, now for the real world

[sonicmerlin]

But you could easily have a company that owns a chunk of spectrum in say a single urban environment deploy a fixed wireless DIDO system. This would provide huge amounts of competition to the duopoly DSL/cable ISPs.

I was skeptical at first

[mcmonkey]

How do you come up with signals that not only constructively and destructively interfere in precisely the right spot in precisely the right way to deliver data to a device, but also for those same signals to simultaneously interfere at other points to deliver different data?

Designing radio signals that will interfere with one another in just the right way takes complex mathematics and careful coordination among the different DIDO transmitters. "The computational requirements are very large, but we solved that by using a cloud server," says Perlman.

Oh! The cloud. I thought he might dodge the question with some hand-waving. But he's got the cloud on it.

Where do I sign up? And how do I make sure the guy sitting next to me isn't stealing my signal?

Re:I was skeptical at first

[powerlord]

Oh! The cloud. I thought he might dodge the question with some hand-waving. But he's got the cloud on it.

Yup. Since a Cloud is just a Virtualized Data Center, there's no reason why it will be a problem in a production deployment ... I mean, we can always just throw another Cloud at the problem.

Where do I sign up?

I'm sure they'll let you know.

And how do I make sure the guy sitting next to me isn't stealing my signal?

Large bladed object (through the interlopers device of course) should remove all worries (and probably give you some nice breathing room at the crowded cafe ... at least till the authorities arrive.

Re:I was skeptical at first

[Chris Pimlott]

That's a good move. Clouds are everywhere, the coverage area will be huge!

Re:I was skeptical at first

[Ungrounded Lightning]

How do you come up with signals that not only constructively and destructively interfere in precisely the right spot in precisely the right way to deliver data to a device, but also for those same signals to simultaneously interfere at other points to deliver different data?

For each of the N partners you compute a signal that puts a null on all the N-1 OTHER partners but not on the partner of interest. That can be heard everywhere except near the other partners - and the one you're talking to is somewhere other than on top of the other N-1. Do this N times and add the results.

This takes a minimum of N separate antennas.

For reception, you use the same NxN amplitude/phase[antenna,partner] matrix to compute N virtual reception patterns, each with a null on all but one of the partners.

I don't buy it.

[Anonymous Coward]

Need to see some numbers and math first.

Re:I don't buy it.

[Lunix Nutcase]

No you don't. The cloud already did all the math for you.

Re: Move Slightly!

[Anonymous Coward]

Damn! You momma is so fat so when your dad fscks her, their wireless connection is broken!

Re:New ? Hardly.

[Mia'cova]

That's just to help the signal. It still shares bandwidth amongst all users. With this, each user can theoretically get full-spectrum downstream. Also different in that it broadcasts from multiple access points which is hardly trivial.

Phased Arrays

[naasking]

In other words, phased arrays. It'd be really cool if he could deploy it for wireless communication though. There's a lot of wasted wireless bandwidth to recoup.

We'll see, but don't get your hopes up.

[Anonymous Coward]

It's not a bad idea to try to resolve multipath inteference with better code separation algorithms (phase orthogonality through increased sampling rates), don't think their claims are valid, or that this method is scalable...

Re:New ? Hardly.

[boss_hog]

Bingo. It sounds like he's trying to take a "cloud-sourced" approach to MIMO, with a little meshiness thrown in for good measure.

Plus I think the whole "support for non-stationary receivers is a huge issue" and "needs to avoid interference that's not of its own making" aspects will make this a non-starter. Good luck getting that spectrum, or finding a big enough group of fixed-wireless customers to make this either useful or profitable.

WiMAX and LTE are already doing MIMO and beamforming (perhaps to varying degrees), so the only thing novel about this is how massively it can fail, and just how smoothly he managed to weave the ever-magical "cloud computing" buzzword into it.

Hmmm...

[vegiVamp]

I can't seem to find any reference to it, but I read about a similar system several years ago, where communications for a submarine would be split up into several waves, which only combine into a useful signal at the point the submarine is supposed to be.

Don't know wether it was an idea or something that was actually implemented, though.

Re:Hmmm...

[Anonymous Coward]

I'm not sure if that's exactly the same. From what I read in the whitepaper, it seems all clients and access points use the same channel. The closest comparison I could make is that it's like differential signalling. Except in differential signalling you send a signal and it's inverse down 2 lines that are parallel the entire run. They both get the same interference and so you can do some comparative math to figure out what the signal really is and isolate the noise.

It seems in DIDO, there's a data center that all APs connect to. Part of their initiation process is to listen for interference from other DIDO APs. This interference is sent to the data center which probably looks through it's transmission logs to determine the identity and relative strength of the interfering AP. Then it does the comparative math on the two separate signals prior to even sending the wave forms to the APs.

I don't think this would help if there was interference on the DIDO channel from a non-DIDO device, though.

Re:New ? Hardly.

[billcopc]

Again, Ciscos can do this. I don't care much for the company, but I've a client with more money than brains and they have a HUGE deployment of these things. The WiFi is actually faster than the wired lan, despite having 300+ clients.

What about secondary interference points?

[Annirak]

Waves don't only interfere constructively at one point. They interfere constructively at many points, to varying degrees. What happens when two devices are using mirrored interference points?

Instead of targeting specific devices, what about dividing the landscape into many physical regions, using constructive interference to cover an area rather than a single device. It would be like space-division multiplexing.

My biggest concern with this tech is not transmission from towers to individual devices, but rather the return call. What are the computational requirements for a receiver using this technology?

Re:What about secondary interference points?

[wcoenen]

This trick only improves bandwidth in one direction. It requires multiple transmitters in different locations, so the phone can't use the same trick to respond.

Re:What about secondary interference points?

[skylerweaver]

According to the whitepaper, the coefficients to weight each transmitter signal to constructively interfere at your location sets up mathematically orthogonal channels (at lease orthogonal to some SNR, with some leakage from other channels depending on the number/location of devices and antennas).

The device can send a signal back which will interfere with other devices, but incoming signals at the antennas can be weighted by the same coefficients (or at least derived from the same) to again cancel all the other signals but your own.

Mathematically, the channel can go both ways with full bandwidth.

Re:What about secondary interference points?

[nzac]

It will use directional transmission not radial waves.

Though if you have a mesh of transmitters you would could use radial waves i would think but the complexity function based on the number of users and nodes would be exponentials within exponentials.

This sounds familiar...

[Anonymous Coward]

Wireless N much?

Re:New ? Hardly.

[alta]

May suck for cell phones, but it'll be great for me. I live in an area where I'm just barely out of range of DSL and cable. Not anything like montana or alaska... Southern alabama, Right between Mobile and pascagoula. I'm currently using 3G wireless from verizon, and it pretty much sucks. 1.1MB down, when it works, the rest of the time, SOL.

My stationary USB card in a 3G router would love to site nice and still for this to work.

Phased array

[Anonymous Coward]

fyi, it's a phased array (http://en.wikipedia.org/wiki/Phased_array)

Like VERY complicated differential signaling?

[Anonymous Coward]

I read the whitepaper. I seems like it's a novel and complicated implementation of differential signalling, the same stuff that prevents interference in Ethernet twisted pairs.

Of course, they need a data center to compute the waveforms. It's much more complex than normal differential signalling. It's not a simple inverse partner signal, it's the background noise of several different signals. But if you know the relative location and strength of each interfering radio, it'd work. I'm surprised nobody thought of this before.

Re:Like VERY complicated differential signaling?

[Anonymous Coward]

I'm surprised nobody thought of this before.

They did, actually. And some test implementations have been done. Don't think it ever went commercial though, because as you note it's extremely complicated to pull off.

Quicktime is not a codec

[Anonymous Coward]

Quicktime is not a codec, it is an extensible proprietary multimedia framework.

bogus marketing hype

[Anonymous Coward]

The "new" technique sounds a lot like multi-user MIMO as used, for example, in LTE and WiMAX. However Mr Perlman makes no acknowledgements of this prior art.

The white paper that he published reads like a marketing brochure; the "technical" bits are a joke. He glosses over the hard parts (e.g. getting accurate channel state information from the devices; the high bandwidth required on the backhaul links from the access points; the incredible amount of processing power required in his "cloud" server farm; etc etc) and makes unsubstantiated claims (e.g. "no upper limits").

And don't get me started on his portrayal of Shannon's Law -- he might "understand" the equation but hasn't a clue how to apply it in a multi-acess system.

I'll believe it when I see it.

Re:Before we get too excited over this ...

[hamburgler007]

What's your point, that the guy is brilliant? Does the fact that quicktime ultimately didn't become the video standard detract the technical accomplishment?

Old news....

[Anonymous Coward]

Got this from steve's twitter feed ages ago... [YAWN]

Re:New ? Hardly.

[Ungrounded Lightning]

802.11n already does this, they call it "beam-forming". Cisco features it in their high-end access points, using multiple antennae to send the same payload but with varying phase shift, which recombine at the receiver to produce a stronger coherent wave.

Which is a variant on "steerable null" - a multi-antenna hack that lets the antennas at a cell site send out beams configured such that, at each active remote device paired with the site, the signals intended for all the OTHER active receivers cancel out. (Ditto with those coming FROM the remote devices to the cell: A combination of the multiple antennas' signals is computed for each remote terminal, in such a way that the signals from all the OTHER remote terminals cancel.)

This one seems to be a 3-D version of the above. Instead of a tight cluster of antennas at a site making beams with nulls pointed at all-but-one of the remote termals, it uses an array of synchronized transmitters to give each partner a signal distribution that has a dead spot on each of the other partners.

Both require (at least) as many cooperating antennas as remote partners, by the way. You can't generate more than one spectrum of signal from each antenna.

Um, MIMO?

[Anonymous Coward]

Multiple antennas at Tx and Rx, MISO (multiple transmit antennas, only a single rx), phased arrays etc, have all been studied extensively for the past 2 decades (and longer in fact). All the upcoming WAN/WLAN technologies have some flavors of this. For example, check 802.16e, LTE, LTE-Advances, .11n. Mutliple-user MIMO and the so-called "Network-MIMO" are also in the works. Note that none of this comes for free, it takes a reasonable density of pilot signals to estimate the channel and will quite likely only work at medium to low user speeds.

Quite wrong to say that this Pearlman guy has suddenly "come up with" this. Looooong list of publications/patents there.

admin costs built into the voice plan

[Chirs]

If admin overhead was the reason, then graphing cost vs data allocation should result in a straight line that crosses the axis at a value equivalent to the overhead. Instead, the higher data rate plans become progressively cheaper even factoring in some constant amount of overhead. For example, my local telco has the following plans for mobile internet:

$15 250MB
$25 1GB
$60 3GB
$75 Unlimited

Strangely, their 4G iPad plans are totally different:
$20 500MB
$35 5GB

Re:admin costs built into the voice plan

[xatm092]

Surely it's because they want you to fork out the extra money for the next one up?

Codecs vs. Containers

[Solar-Powered Rocket]

I love how the summary introduces him as the "inventor of the QuickTime codec".

A common enough confusion I suspect. To be pedantic: QuickTime is a media container, not a codec. It's similar to the way that AVI and OGG aren't codecs. They're containers for stuff like MP4 (confusingly sometimes also the name of a container format), Vorbis (the codec behind most Ogg audio files), or Mp3.

We're Toast

[umbrellasd]

This is the same kind of technology used to take 3 innocuous beams of light and explode your head at the point where they cross-over, and my phone has alarmingly accurate location information these days...

I can see it now. Sprint hires hacker to hack the T-mobile phone network and in a single keystroke explode their customer base; other networks follow moments later. The first and final act in what is later to be known as the Carrier Wars.

Re:We're Toast

[Yamioni]

Seems kind of dumb to kill all the customers of a competing company. Those are paying customers you'd much rather have for yourself; dead men pay no bills. I'd hire the hacker to limit the head 'splosions to the people actually running the competitor. Then when the competitor collapses due to having no more employees, where are all the people needing cell service going to turn? That's right.

So no cause for alarm here, people... Unless you happen to work for a telco that is...

DECCA

[Anonymous Coward]

DECCA, used for maritime navigation. Been there, done that, nothing to see here, move along, move along....

Perlman says it's not beam-forming

[Namarrgon]

In his June 4 presentation he states that it's "not beam-forming". He doesn't say much about what it is, though.

His white paper (PDF) gives a bit more detail, though still not much. It sounds akin to MIMO, but instead of phase-aligning multiple signals to increase the strength (i.e. beam-forming), the antennae are more widely distributed, and complex-formed signals are broadcast from each antenna in careful sync, so that they interfere at each receiver to produce the desired signal.

Can't beat Shannon this way.

[goodmanj]

This is cute, but it won't let you beat information theory limits on signal bandwidth. After all, no matter how these signals are intended to be combined through spatial interference, each antenna is emitting a signal which varies only in time. So the more signals you try to pack into one antenna's output, the more those signals project onto one another ("overlap"), and so the more they get mixed together. From each antenna's perspective, this is just a baroque form of time-domain multiplexing (TDMA) -- more accurately it's phase-domain multiplexing (PDMA) -- and all the usual rules about maximum bandwidth per user per antenna still apply.

Re:New ? Hardly.

[black soap]

It sounds like they demonstrated really-high bandwidth to a single specific user, but how well does it provide high-bandwidth for multiple users? Wouldn't it get calculation-intensive, as each signal is modulating the others?

Inventor of *the* QuickTime codec? Really?

[Anonymous Coward]

QuickTime is not a codec. QuickTime is a container format (and an API). Road Pizza was a very early Apple codec available in QuickTime. It may seem pedantic to make this distinction but it's one I would hope Slashdot's editorial standards would adhere to.