Neat. The phenomenon is called Rayleigh fading, and is well known, but I've never seen it mapped as clearly. Rayleigh fading is often viewed statistically, and some article even call it "random". As shown here, it's not random at all; it's deterministic and repeatable. It's determined by the position of transmitter, receiver, and reflective and absorbent objects in the environment.
This is why most WiFi devices have two antennas and a diversity receiver.
Spread-spectrum systems are resistant to Rayleigh fading, because multipath nulls are frequency-specific; the two paths have to be half a wavelength different in length to get cancellation. So it seems surprising at first to see this with WiFi, which is a spread-spectrum system.
There's a reason for that. Cancellation becomes less of a problem as you get further from a spread spectrum transmitter, because the phase difference is related to the product of the frequency change times the number of wavelengths along the path. (This really needs pictures.) Once you're far enough away that the spreading moves the nulls at least a half wavelength, the effect is that a fixed percentage of data is lost, and error correction deals with that. In this demo, transmitter and receiver are very close, so the nulls are strong.
That is slick. The wavelength of 2.4Ghz radio being 125 mm and 5Ghz being 62.5mm it looks like he is imaging standing waves in the 5Ghz spectrum. (if the cube is 360mm x 360mm it would have ~6 waves of 5Ghz and ~3 waves of 2.4Ghz RF energy. There is an experiment to see the standing waves in a microwave using chocolate - http://morningcoffeephysics.com/measuring-the-speed-of-light... which demonstrates the same sort of thing, but frankly I think this is much cooler.
He does mention that explanation for explaining the distance between "features" in his visualization, however this does not explain their particular shape, nor their general irregularity. It would be nice to see this repeated outdoors in the desert to reduce reflections and possibly interference.
Reminds me of another project visualizing wifi signals:
> This project explores the invisible terrain of WiFi networks in urban spaces by light painting signal strength in long-exposure photographs.
> A four-metre long measuring rod with 80 points of light reveals cross-sections through WiFi networks using a photographic technique called light-painting.
That is interesting also, but it is 1D only, the light stick's total illuminated length is signal strength. If they re-did that project using RGB LEDs and each LED colored from Red to Blue based on the signal strength at the LED then you would get a 2D plot, And of course if did as the author and photograph an LED at intervals in a 3D space you could do the 3D plotting.
The whole thing has my head buzzing with ideas. Imagine a drone which has a string of RGB LEDs hanging from it making such plot using GPS/6DOF IMU data to allow a computer to reconstruct the scene. Could be very cool indeed.
I was thinking the same about drones. Take a drone and let it automatically cover an area you're interested in. Maybe following a certain pattern or just random movement.
Because terrific work causes us to think of additional questions, I'd add: The only thing I found missing is that there is more than one polarization. So to accurately map the field, you'd need to perform multiple scans, with the receiving antenna rotated 90 degrees for each one.
Another question: when he uses the CNC to move the sensor around, isn't he also changing the reflective surfaces in the environment? Would this not produce inaccuracies?
Awesome. I'm not sure how to make this argument precise, but I think you could sample the intensity function only every 12.5cm (wavelength at 2.4 Ghz), since waves of a given wavelength are restricted to resonate and interfere at small multiples of it's wavelength.
For reconstruction there should be some ideal interpolation, maybe a 3d sync filter?
Another random interesting fact: if you capture not only the intensity but also phase of only a 2D slice, and have some environmental information (i.e. know how the signal propagates in your room), you can reconstruct the full 3D intensity.
A naive application of Nyquist sampling theorem will say you have to sample at twice the frequency of whatever you're trying to reconstruct - so for 2.4Ghz you need to sample at at least every 6.25cm. The reason why you can't just use the base frequency is cause... well, for example, imagine you have the worst luck ever, and only ever sampled the nodes of the standing wave - it would just look like no signal.
In practice... I have no idea. RF is too much for me.
Signal variance is nothing new. I remember personally noticing it in late 90's that with my GSM (900MHz) phone, moving it on table just for about 5 centimeters could bring it down from full signal to no reception at all. Of course with analog radios you also notice how easily signal changes with location. If you've been ever playing with TV antenna in in bad reception and so on. Moving your hand in other room might block TV signal or make it crystal clear even if you would make there's no connection what so ever. With 2.4GHz people often forget that interference from other sources can significantly contribute. So signal quality and signal strength aren't same thing at all. Getting to the root all these things require professional, which I'm not. So one type of measurement defined as "signal strength" probably misleads you badly. Is it a good idea to select a wifi channel that doesn't have any other wifi boxes? Well, the reason might be that the channel is totally overpowered by local wireless CCT or phones. That's the reason why nobody's using it for WiFi and then you think it's a great idea to select a free Wifi channel?
Radio stuff is (truly) really tricky. With higher frequencies it's just like light. Why some things are in shadows and some things are well lit?
I'm in a relatively large appartment building.. I can see about 20 other wifi options in my area on my laptop... fortunately, my signal quality is significantly better (higher end asus, with shibby tomato) spent a fair amount of time tweaking the settings a bit that seem to work very well for me... I can keep signal on my phone across the parking lot (I'm facing the pool, which is adjacent to outside parking) ... no wifi really gets into the garage though, but cell phone signal does...
It's really wild how a foot or two difference could make all the difference in the world...
Well, yes to both questions, but on a macroscopic scale the wave equation can explain the collective effect of zillions of photons to more accuracy than you'll ever need.
I'm curious if combining this with Google's Project Tango for accurate 3D pose estimation would be useful for doing a handheld 3D mapping of an area. Add an additional camera to the tablet, have it point at an array of these sensors, and use it to sweep out regions of space quickly.
Pretty cool! Would be cool to add some sort of position sensor on, I guess GPS wouldn't be ideal though. Maybe hack a roomba to map signal strength in your house!
There was a project several years ago to map a University's wifi using a self driving RC car with a wifi detector on it. The goal was to identify weak spots in the signal. I can't remember the name of the project and its not popping up on google, anyone else remember this?
This is brilliant, I wish it'd get more upvotes and attention. Wireless waves are such nebulous, hard to understand things. It's great to see a visual representation!
The most top ranked comment says; "PATENT THIS MOTHERFUCKER NOW, DO IT BEFORE APPLE FIND OUT AND STEAL YOUR IDEA." Hope neither this will get patented or Apple will steal it.
This is why most WiFi devices have two antennas and a diversity receiver.
Spread-spectrum systems are resistant to Rayleigh fading, because multipath nulls are frequency-specific; the two paths have to be half a wavelength different in length to get cancellation. So it seems surprising at first to see this with WiFi, which is a spread-spectrum system.
There's a reason for that. Cancellation becomes less of a problem as you get further from a spread spectrum transmitter, because the phase difference is related to the product of the frequency change times the number of wavelengths along the path. (This really needs pictures.) Once you're far enough away that the spreading moves the nulls at least a half wavelength, the effect is that a fixed percentage of data is lost, and error correction deals with that. In this demo, transmitter and receiver are very close, so the nulls are strong.