A few Wi-Fi transmitters and your house becomes a coprocessor | Hacker News Reader

A few Wi-Fi transmitters and your house becomes a coprocessor

arstechnica.com

78 pointsrbanffy15 comments

Tangent: What is the term for these sorts of introductions?:

> In computing, we have pretty much come to accept the idea that smaller is better. The smaller the features on a chip, the more transistors we can place, the more computation can be done.

This doesn't actually help the reader understand the contents of the article. At best, it tries to make the reader care about the article by drawing a vacuous comparison to a generic piece of conventional wisdom.

There's a similar thing going on when I try to Google for recommendations/reviews for, say, an exercise app. Very often, the article will begin with just absolutely useless words like

> Research shows that people who exercise are happier and healthier. But this fast-paced modern life doesn't always make it easy! There's just so many things to get done. How can we make sure we're exercising as efficiently as possible? Workout apps! But which ones? Let's take a look at some of the different workout apps available in the app store.

Is this sort of filler material being added because writers are getting paid by the word (honest question), or is it actually demanded by the editors of the websites because it attracts more readers? Or is it some sort of SEO thing?

Just trying to understand this phenomenon.

Yeah, I think the entire article is written poorly and quite incoherent

Perhaps the thesis, antithesis, synthesis triad? Also known as the Hegelian triad.

For the linked article, thesis = "In computing, we have pretty much come to accept the idea that smaller is better." Antithesis = "Researchers are proposing an idea to make your computer bigger." Synthesis = "They want to turn your entire house into a co-processor."

I've seen this and similar story templates show up in articles and books on writing, especially related to writing for a mass audience. They are supposed to produce narrative tension and improve reader engagement. Results vary.

https://en.wikipedia.org/wiki/Thesis,_antithesis,_synthesis

Thanks, that's useful.

You've just put words to something I had unconsciously noticed. Particular types of site I know to scroll about 1/3 down the page to get the actual information

I remember always being tempted to write these kinds of introductions in my high school English classes. Nobody bothered to tell me it was wrong until a college writing class. I think some teachers even thought it was good.

I'm not a physicist like the Ars writer, but I always kinda though that computation requires energy consumption and non-linear elements to interact with a signal. The only energy source in this experiment are the Wifi transmitters and I don't see any nonlinear elements.

The paper abstract seems to define 'computation' as performing a Fourier transform like an optical lens (I'm imagining a prism separating the wavelengths). It also says "We can exploit the linearity of the computation operation Y = GX in which the entries of Y are independent from each other."

So I don't see how this meets any useful definition of "coprocessor". Can you use it to perform a NAND operation?

No it does not, it is the Ars journalist that write about coprocessor, the Arxiv article is about infering the matrix G, nothing else.

They do it in a quite clever way.

You have to be careful about what you mean by "linear"—linear programs are P-complete, and all binary operations are linear over F_2 (including NAND, etc), since all functions are polynomials over finite fields and the largest polynomial over F_2 is of degree 2-1=1. There are many linear elements which can generate arbitrary circuits and hence can (in principle!) implement any finite circuit (e.g., processor).

I do agree, though, that in this case it's unclear how the 'coprocessor' connection arises from the original statement given in the paper.

"Linear in F_EM" :-)

This seems similar to what Artemis Networks has been working on, although they don't use the waves for computation but communication.

https://www.artemis.com

Hi, I admit that when I read your comment, I wondered what it has to do with the Ars article. In fact after thinking a bit I think you may have better understood the Arxiv article than the guy at Ars.

Indeed the point of the Arxiv article is not about doing computations, it is about Time Reversal/Wave front shaping.

The original contribution of the authors seems to be in infering the matrix to do beam forming simply from CSI in any random environment.

Citation: "We have demonstrated experimentally that any disordered medium can be employed as reconfigurable analogue computation unit, subject to appropriate wavefront shaping"

> beam forming simply from CSI in any random environment

Isn't this basically holography, just using a few discrete antennae rather than an (approximately) continuous film medium?

I have a good understanding of Wavefront shaping but holography is out of my comfort area.

I DDG a bit and found that "The difference between a hologram and a conventional photograph is that conventional photography records intensity of light while a hologram records both intensity and phase."

So maybe...

Holography is about encoding a high dimensional (3d) structure in a lower (2d) dimensional plane, that this done by cleverly modifying a wavefront is mostly to do with light being waves.

Reply on news.ycombinator.com