Slashdot Mirror


Researchers Want To Turn Your Entire House Into a Co-Processor Using the Local Wi-Fi Signal (arstechnica.com)

An anonymous reader shares an excerpt from a report via Ars Technica: Researchers are proposing an idea to make your computer bigger. They are suggesting an extreme and awesome form of co-processing. They want to turn your entire house into a co-processor using the local Wi-Fi signal. Why, you may be asking, do we even want to do this in the first place? The real answer is to see if we can. But the answer given to funding agencies is thermal management. In a modern processor, if all the transistors were working all the time, it would be impossible to keep the chip cool. Instead, portions of the chip are put to sleep, even if that might mean slowing up a computation. But if, like we do with video cards, we farm out a large portion of certain calculations to a separate device, we might be able to make better use of the available silicon.

So, how do you compute with Wi-Fi in your bedroom? The basic premise is that waves already perform computations as they mix with each other, it's just that those computations are random unless we make some effort to control them. When two waves overlap, we measure the combination of the two: the amplitude of one wave is added to the amplitude of the other. Depending on the history of the two waves, one may have a negative amplitude, while the other may have a positive amplitude, allowing for simple computation. The idea here is to control the path that each wave takes so that, when they're added together, they perform the exact computation that we want them to. The classic example is the Fourier transform. A Fourier transform takes an object and breaks it down into a set of waves. If these waves are added together, the object is rebuilt. You can see an example of this in the animation here.

53 of 102 comments (clear)

  1. For the record... by Anonymous Coward · · Score: 3, Insightful

    Whoever wrote this proposal is dumb as fuck

    1. Re:For the record... by Darinbob · · Score: 1

      I learned from the article that "wi-fi" is a synonym for radio waves. I feel smarter already.

    2. Re:For the record... by RespekMyAthorati · · Score: 1

      Actually, the Ars Technica "journalist" is dumb as fuck and full of shit.

      The actual paper is about replacing expensive meta-materials with less expensive
      common materials in wave transmission research.

      That's all. The paper says fuck all about doing computing with household routers.

  2. What month is it? by Kohath · · Score: 2

    It's supposed to be April 1st. That's why they call it "April Fools". You can't run this stuff the first of every month. Only April.

    1. Re:What month is it? by bugs2squash · · Score: 1

      It may have been posted on May 1st here, but maybe BeauHD is somewhere that is in the timezone UTC-720 where it is April 1. Either that or 482 billion miles away.

      --
      Nullius in verba
  3. Need a little popcorn by LynnwoodRooster · · Score: 2

    So I can eat it when people are debating about the microwave screwing up their coprocessor - or not...

    --
    Browsing at +1 - no ACs, I ignore their posts. So refreshing!
  4. FARMING... F' off by wolfheart111 · · Score: 1

    enough said

    --
    [($)]
  5. Ho Hum by az-saguaro · · Score: 2

    The end of the first paragraph in the post states "But if, like we do with video cards, we farm out a large portion of certain calculations to a separate device, we might be able to make better use of the available silicon." What I was expecting at that point was an idea in distributed computing. Your processor idles down or else is overwhelmed by a high bandwidth task, so it offloads portions of the job to your refrigerator, cell phone, tablet, washing machine, other desktops. you car, etc. - anything that can be accessed by IoT or peer-to-peer networking or whatever. That makes sense, kind of, if you are doing supercomputing tasks or cryptocurrency mining at home.

    However, this turns out to be an idea about using multiple wifi antennas to create interference patterns from which FFT's can decode the dataset - kind of a wifi holography. The problem is that anyone who moves the desk or couch, leaves their bicycle in the room, puts a metal kitchen bowl on a table, turns on a laptop, has their Aunt Tillie sitting in the wrong place, or even just wears a tinfoil hat to tune in or tune out other strange ideas will trash the computation.

    It used to be that with relatively few TV channels and finite numbers of radio stations and newspapers, that news could be filtered or curated to things of genuine significance. Now, with seemingly unlimited media outlets, any idea or premature utterance can become "news". A century ago, we fantasized about going to the moon, and then it happened. Just half a century ago, we fantasized about handheld computers and wrist watch televisions, and then it happened. So, ideas can come to fruition, so we cannot be too dismissive. This one however seems to be a big "so what", and "what the hell would one use it for?"

    1. Re:Ho Hum by msauve · · Score: 1

      "What I was expecting at that point..."

      If you had kept reading, your expectations would have become much, much, lower. The whole thing seems to have been written by a middle-schooler armed with a random punctuation generator.

      --
      "National Security is the chief cause of national insecurity." - Celine's First Law
    2. Re:Ho Hum by coofercat · · Score: 1

      I'm also wondering about the real-world use of all this. Adding waveforms together is really just that - adding. It seems that asking some waves to do some algebra or trig seems like a bit too much to ask. So I wonder how much processing you could really offload with this...?

    3. Re:Ho Hum by shaitand · · Score: 1

      "Adding waveforms together is really just that - adding."

      There are a lot of reasons this is a terrible plan but that isn't one of them. It has been proven that every logical operation can be performed with negated binary addition. It may not be the most efficient way but it can be done.

      https://www.allaboutcircuits.com/technical-articles/universal-logic-gates/

    4. Re:Ho Hum by rubycodez · · Score: 1

      every logical operation can also be performed with an abacus. So what? article proposes a similarly stupid thing, and proposing doing calculations that way is stupid, slow, inefficient and pointless

    5. Re:Ho Hum by shaitand · · Score: 1

      So... as I said in the comment you responded to, this is a terrible idea but the flaw is not as suggested by the GP that the modification to waveforms is addition.

      I said "... this is a terrible plan... It may not be the most efficient way" You seem to be rephrasing the same sentiment I express but with a tone implying it is somehow an argument against what I said.

      By the way, if we are talking about using universal logic operators rather than this wifi radio proposal on the whole you should be aware that for the most part the fastest general purpose processors are following a similar path at the transistor level.

  6. Imagine A... by 0100010001010011 · · Score: 2

    beowulf cluster. Seriously Slashdot, you're slipping.

    So basically distcc, icecc, ore any of a number of other tools?

    1. Re:Imagine A... by SeaFox · · Score: 2

      You mean a Beowulf neighborhood?

    2. Re:Imagine A... by Tablizer · · Score: 1

      In Soviet wi-fi house, you are the Beowulf Cluster.

  7. Geniuses. The people who funded it, however by raymorris · · Score: 4, Insightful

    The people who wrote this proposal got tax payers to pay their next two years expenses while they fart around with a wifi router.

    The bureaucrat who approved it and the tax payers who are funding it might be dumb as fuck, however.

    1. Re:Geniuses. The people who funded it, however by thegarbz · · Score: 4, Interesting

      And just like all other science which sounds like a bullshit waste of time when it is done it may also yield some new understandings of how interacting radio signals can be used for practical purposes.

      Sidenote: Does anyone know of a news for nerds site? I'm looking for a site which specialises in technology and has an interested readership. All I can seem to find is a bunch of negative luddites who are more interested in blaming the entire world on governments.

    2. Re:Geniuses. The people who funded it, however by Anonymous Coward · · Score: 1

      I will just install a quantum computer emulator on my microwave and be done with it.. then it can send its magic quantum waves directly to my computer and it can turn it into a rainbow colored unicorn that resembles a my little pony.

    3. Re:Geniuses. The people who funded it, however by jpaine619 · · Score: 1

      Yeah! It's not like anything has ever come of research into radio waves. It's wasted money!

      Radar

      FM Radio

      TV

      HAM

      P.S. You are an idiot.

    4. Re: Geniuses. The people who funded it, however by kenh · · Score: 3, Insightful

      And just like all other science which sounds like a bullshit waste of time when it is done it may also yield some new understandings of how interacting radio signals can be used for practical purposes.

      Bullshit.

      What this article claims is that complex calculations (like fourier transformations) can be computed, and their results captured, based on carefully controlling the interaction of WiFi (or other RF) signals. I suppose it could be argued that this could create a form of analog computer, but the article implies that any calculation might be fair game, up to (I assume) mining crypto currency.

      Bottom line, they seem to be proposing that we take several small processors, use them to carefully modulate/regulate the RF emissions of several routers to perform otherwise trivial calculations?

      Bullshit.

      --
      Ken
    5. Re: Geniuses. The people who funded it, however by kenh · · Score: 2

      They propose using several computers to regulate RF emissions and an additional computer to detect/analyze the resulting interference to perform a calculation that could be performed trivially by any of the individual computers in the experiment.

      It's like you've never heard of Rube Goldberg.

      --
      Ken
    6. Re: Geniuses. The people who funded it, however by edris90 · · Score: 1

      Well whatever entity is claiming authoritative control is responsible and liable for all outcomes in their declared territory. Such as a trade-off for controlling all is now your fault. However you're right everything has gone political lately that's probably because from people noticing the all the tech is pointless and politics and business interests sabotage implementing it properly in basic infrastructure.

    7. Re:Geniuses. The people who funded it, however by Moof123 · · Score: 2

      As a RF/Microwave design engineer of 20 years I found the summary and article to be pretty incoherent.

      As best I could discern that gist of things through the mangled technobabble and hype the argument is that if you can fully map an interference pattern between two sources you can then perform some calculations by measuring the resulting new interference pattern between two new signals. I sort of see what they are going after, but many details are lacking as to who you make this work.

      Caveats:
      1) Both signals must be coherent, i.e. same exact frequency and phase locked. WiFi signals are usually not phase locked, and channels are chosen arbitrarily.
      2) Measuring the result requires a bunch of sensitive RF downconverters. These burn power. The results are a type of analog computer, so don't expect more than a couple digits of accuracy.
      3) Anything that changes the interference pattern ruins the characterization. So if you move in your room, the characterization has to be re-done. Super useful...
      4) With a handful of receivers in a room the rate of computation would be excruciatingly slow, making this a pointless party trick that only the nerdiest of geeks would ever appreciate.

    8. Re: Geniuses. The people who funded it, however by kenh · · Score: 1

      Bottom line, he seems to be proposing that we take a bunch of vacuum tubes and wires, use them to carefully modulate/regulate the voltages in a bunch of other wires and tubes to perform otherwise trivial calculations?

      Taking one thing to make another thing is called 'invention', taking something and and making it do what it was already designed to do is also called 'invention', but only if you are a student named Ahmed, live in Texas and you 'invent' a digital clock by taking it out of it's case and putting it in a pencil box.

      --
      Ken
    9. Re:Geniuses. The people who funded it, however by rubycodez · · Score: 1

      Wrong, there is nothing new to be learned about "interacting radio waves' for the frequencies in question in air. Done deal a century ago, study electromagnetics and learn why.

  8. Great idea! by no-body · · Score: 1

    Your cells and DNA will just love getting bathed in constant WIFI stimulation...

  9. Re:Ok, it's possible. BUT WHY?! by AHuxley · · Score: 1

    People want 4K. But don't have the skills to review and then buy a new GPU and CPU to support that.
    So the small trendy low power computer can pass the complex math to other devices and they can push the results back.
    A huge increase in real time CPU and GPU power with no lag is the result.
    Modern computer math magically no longer needs the CPU and GPU to be in the same part of the house.
    The networked refrigerator, toaster and air conditioner CPU will all get the complex GPU math done with no lag.
    Dropping frames again with that more advanced computer game? Turn on the networked washing machine and do some laundry to add more CPU power.

    --
    Domestic spying is now "Benign Information Gathering"
  10. Does not compute by gopla · · Score: 1

    If I were the funding agency, this would be reected instantly. You can heat your house better by doing random switchings of transistors rather than this round about way. It would have been better if these researchers were honest in their request. I am not even sure they have any case or even trivial computation like Fourier transform through their method.

  11. Fake news by Anonymous Coward · · Score: 1

    The arxiv abstract has very little to do with what the article blurb says.

  12. Re: It could be useful for encryption by Miamicanes · · Score: 1

    We blew past 128-bit computing (or at least, 128-bit registers with 64-bit data bus and 48+ bit address bus) YEARS ago (ok, technically, the general-purpose registers RAX through R15 are "only" 64-bit... but the MMX registers ain't just for graphics anymore). Sandy Bridge bumped us up to 256-bit registers. Knights Landing raised that to 512 bit registers (AMD's best processors are comparable, and I think they doubled the number of huge registers from 8 to 16).

  13. magical thinking by psb777 · · Score: 2

    Magical thinking. Perhaps they could sell a skull cap for my cat? Wouldn't want any processing power going to waste.

    --
    Paul Beardsell
  14. This will be used for mining by xack · · Score: 1

    The new law is any advance in computing power will be used for crypto mining, until the difficulty adjusts bringing back an equilibrium.

  15. Re:Ok, it's possible. BUT WHY?! by gl4ss · · Score: 2

    People want 4K. But don't have the skills to review and then buy a new GPU and CPU to support that.
    So the small trendy low power computer can pass the complex math to other devices and they can push the results back.
    A huge increase in real time CPU and GPU power with no lag is the result.
    Modern computer math magically no longer needs the CPU and GPU to be in the same part of the house.
    The networked refrigerator, toaster and air conditioner CPU will all get the complex GPU math done with no lag.
    Dropping frames again with that more advanced computer game? Turn on the networked washing machine and do some laundry to add more CPU power.

    don't be stupid. read this proposal again. this is way more sillier than that.

    this is basically turning your house, with use of wifi signals, into a fft analysis machine or something similar, like a huge mechanical computing device except with wifi instead of levers.

    100 bucks they got the idea from engineering guy demonstrating the fourier calculation mechanical machine lately on youtube.

    the idea is dumb as fuck. turning your sink into a computer makes WAY MORE FUCKING SENSE than this.

    --
    world was created 5 seconds before this post as it is.
  16. Let me know if you find one. Tampermonkey by raymorris · · Score: 3, Interesting

    Let me know if you find such a site.

    I COULD extend my old Tampermonkey script to hide all of thr annoying people, or only show comments from people in my whitelist.

    I made it years ago to make that ADP or whatever guy disappear from my browser. That guy who could never understand why in 1982 the world switched from hosts files to DNS. Seen that dude lately? I vanished him with Tampermonkey about three or four years ago.

  17. Gee. They discovered ANALOG COMPUTERS... by Anonymous Coward · · Score: 2, Insightful

    Analog computation is always faster than digital...

    The problem though, is calibration. Analog computers precision always drift and requires yet more analog computing to try and compensate... The end result is about 4 digits of precision, on average...

    Slightly better than a slide rule...

  18. now crowd funding an improved v.2.0 by clangerbanger · · Score: 1

    combined with rotational polarization, allowing for a more thorough fourier dissociation.

  19. Um, what?

  20. Bricked device by iTrawl · · Score: 1

    I guess in this particular case "being bricked" is a good thing for the device.

    --
    "Everybody's naked underneath" -- The Doctor
  21. Because... by kenh · · Score: 2

    Why, you may be asking, do we even want to do this in the first place? The real answer is to see if we can. But the answer given to funding agencies is thermal management. In a modern processor, if all the transistors were working all the time, it would be impossible to keep the chip cool.

    Because there is simply no way to cool a modern CPU with it's millions of active transistors!

    Apparently this research was approved by people that never heard of thermal paste and cpu fans.

    --
    Ken
    1. Re:Because... by drinkypoo · · Score: 1

      Because there is simply no way to cool a modern CPU with it's millions of active transistors!
      Apparently this research was approved by people that never heard of thermal paste and cpu fans.

      Each CPU core is broken up into a number of functional units. Different CISC instructions are broken down into multiple micro-operations, which are executed on functional units. Each core can handle one or two contexts at once, meaning that it's working on retiring one or two instructions at a time, which in turn means that it's got one or two functional units active at a time. Modern transistors are very good both at resisting and conducting, but they still have a state in between where they are doing both, and during that time they heat up. Therefore, when a functional unit is not actually in use, it generates basically no heat, but when it is actually being used it is switching on and off, and heat is being generated as energy is lost to resistance.

      If all the functional units were to be activated at once, the processor would overheat in short order. As it is, many if not most modern desktop CPUs will thermally throttle when used for heaving number-crunching.

      --
      "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
  22. Order of Operations by NEDHead · · Score: 3, Funny

    Honey, could you move the couch an inch to the right? I'm working on my perpetual motion machine calculations. I think I almost have it!

  23. "Waves already perform computations as they mix" by Anonymous Coward · · Score: 1

    Uh, no. No, that's not it at all, electromagnetic waves don't interact with each other. When they collide with matter the sum of their energy is imparted to the thing they're colliding with, with the sum being affected both by the amplitude and phase of the waves relative to each other.

    Ignore the sensational Ars article (which is kind of shocking in itself) and look at the paper. The actual proposal is less "use your house as a coprocessor!" and more, "one of the limiting factors in optical computing research is that it requires the use of difficult-to-manufacture, very high tolerance metamaterials which don't scale well, we think that we can turn any non-uniform material into a reconfigurable computation unit. As a proof of concept, we made some reflectors with which we were able to perform a 16x16 complex value operation using conventional wifi signals as a source."

    In simplest terms, the paper says, "we think we can do optical computing research without micro-fabricated meta-materials, and our proof of concept made out of consumer junk suggests that we might be right."

  24. Except this is not remotely new: homepod by goombah99 · · Score: 4, Interesting

    People have been farting around with optical processing ever since the laser was invented. In the 1980 JPL used optical correlators for image processing. Faster than any computer back in the 1980s.

    These days it's used routinely to have wave pattern defined sub-cells on cellular towers. That's not the same thing as beam steering because it's relying on reflections.

    Apple's homepod sounds better than one can imagine because that's what it is doing too.

    And of course seeing around corners with scattered light is still an active area.

    This particular application is utterly moronic for the simple reason that to do any of the above you need a lot of antennas, There just are not a lot of antennas in a house

    --
    Some drink at the fountain of knowledge. Others just gargle.
    1. Re:Except this is not remotely new: homepod by shaitand · · Score: 1

      Most wifi routers come with multiple antennas onboard now. What makes this moronic in my opinion is that the transistor count to handle all that waveform lifting is going to exceed the count to simply perform the operations. Not to mention the RF interference this system is going to cause.

    2. Re:Except this is not remotely new: homepod by thegarbz · · Score: 1

      This particular application is utterly moronic

      ALL particular applications are utterly moronic in the lab. It's the underlying principles that are discovered that end up forming the most benefits.

    3. Re:Except this is not remotely new: homepod by shaitand · · Score: 1

      "I don't know about that. Lots of places have wifi right now and have for over the past decade."

      And it does cause interference already.

      "Going from random to not random isn't going to change anything related to how interference happens."

      Yes it is, it's going to change the density of transmission. Random means there is no consistent loss. It's like an analog tv broadcast, random noise degrades the picture but it is scattered throughout the image causing fuzz that has to be extremely severe to prevent you from recognizing the image and content despite the errors. What is missing in one frame will be present in the next allowing you to deduce across a sequence what was missing. Not random means strong coherent noise, instead of fuzz that shifts from frame to frame you will have consistent errors that are targeted. It will be more like errors in a digital transmission where you lose entire frames or have solid black bars or analog where you have a constant line of distortion through the image. The total amount of interference might be the same but those errors will be concentrated in more specific ways.

      The more coherent and not random data on a transmission the more it interferes with other signals which is why HAMs have limitation from the FCC on data rates for their transmissions. A sparse 100mW signal causes less interference than a dense one. The more computation you perform the more dense the signal and digital processing is extremely inefficient super dense micro-computation (often using tens or hundreds of thousands of tiny universal logical operations to execute a specific macro routine with a couple dozen potential outputs).

      "The signal strength will be the same, the frequency range will be the same, and the pattern encoded into it will be the same."

      The pattern encoded into it will not be the same otherwise you can't set up and read the result of a specific computation. Without changing the pattern you'd be reading the result of random computations with no relevance to the output you need and you could that with a recording of a waveform and skip the real antennas and wifi.

  25. Re:They want a lot by omnichad · · Score: 1

    Maybe we can replace the heating elements on our toasters with old Pentium 4's and when we want hot bread we can join a render farm.

  26. Re:"Waves already perform computations as they mix by omnichad · · Score: 1

    Wow, that's nothing the same as the summary or headline at all. But at least it sounds actually useful.

  27. Re:New New CPU Powered by Pied Pier by omnichad · · Score: 1

    Standing out along the ocean wearing a Waldo costume?

  28. Almost ... by PPH · · Score: 1

    ... had the solution for the Grand Unified Theory. And then my daughter turned on her hair dryer.

    --
    Have gnu, will travel.
  29. WTFDIJR? by KlomDark · · Score: 1

    What the fuck did I just read? What a load of ignorant crap.

  30. Re:Ok, it's possible. BUT WHY?! by KlomDark · · Score: 1

    Free dumb!