> so it can go through walls
60 GHz RF doesn't really pass through anything, which is the entire reason that the frequency is used for radar. A radar needs powerful reflections to detect things. Penetration impedes its operation, and would be akin to a camera that is out of focus.
I'm also unsure if the resolution of this chip is noteworthy.
For 3$ https://fr.aliexpress.com/item/32786483344.html a 6Ghz radar that goes 12meters using 30mA, so my range claims are not that silly.
When there is no obstacle 6Ghz or 60Ghz doesn't matter.
Power amplifier efficiency for 6GHz and 60GHz is not the same-- not even close.
> For 3$ https://fr.aliexpress.com/item/32786483344.html a 6Ghz radar that goes 12meters using 30mA, so my range claims are not that silly.
This is a deflection and ridiculous. We're talking about spotting e.g. gestures. It falls apart with distance because of both angular resolution and inverse-fourth power.
I've been involved in the design of mmwave radars. If it was easy to spot and precisely track small objects at 10m, we'd be doing it...
That's not the claim I'm making. I agree that this chip won't do the gesture recognition at 10m, but I'm quite convinced that it can pick up human movement signal if they try to do so.
>I've been involved in the design of mmwave radars
I don't have this level of expertise. But I'll be really surprised if we couldn't reach the same levels of amplification. Gaining 12 dB will double the range. We can extend the antenna array or use more expensive low noise amplifiers. For cars there are 30Ghz Doppler radar, and the distance can go at least 50m.
From the referenced paper ( A Highly Integrated 60 GHz 6-Channel Transceiver With Antenna in Package for Smart Sensing and Short-Range Communications https://sci-hub.tw/https://doi.org/10.1109/JSSC.2016.2585621 ) : "In this work a 60 GHz 4-channel receiver 2-channel transmitter packaged chip targeting high resolution sensing systems and capable of supporting large bandwidth communication channels is presented. The SiGe technology used offers a low 1/f noise which is essential to the functionality of the chip in frequency modulated continues wave systems (FMCW)and Doppler radar with a sensing range below 10 m."
"While we have not explored this in-depth we would like to highlight the similarities to re-cent work exploiting FMCW RF technology to coarsely ‘image’ users through a wall [1]"
Your first quote says "sensing range below 10m".
Yes, it is possible to make long range 60GHz systems-- largely through antenna gain and lenses. Yes, we could build an entirely different radar to track peoples' gross movement--- and could have 20 years ago, too-- but that has largely nothing to do with the original system or your original claim.
If I wanted to image people at low resolution through a wall, 60GHz is about the last thing I'd pick. Drywall alone has an attenuation of about 3 dB/cm, and remember we have to cross the wall twice. You run out of loss budget really quick. Suggesting one can go 10x as far (10000x power alone) and through walls is... creative.
If you want to track people through a wall, use UHF. It works pretty well and is pretty easy.
> Suggesting one can go 10x as far (10000x power alone) and through walls is... creative.
I was going for conservative. Typically thinking about open-space work environment. The device doesn't need line of sight like a camera would, it can be in your coworker pocket and listening to you.
They are well founded and doing their own radio chips, I give them the credit they deserve and assume they can replicate a somewhat similar technology (6Ghz radar).
It is better to use a logarithmic scale. The 10000x power (40dB gain) doesn't mean you will need to emit 10000x more power, in practice you will amplify the signal, by focusing the beam via antenna design, and amplifying which theoretically you can do relatively easily as long as you are above the Cosmic Microwave Background Radiation noise. Then you need to do some trickery to trade bandwidth for signal strength (which they do : FMCW). Then you still can integrate over time. A radar typically scan its whole surrounding, but if instead you chose to focus it at one place you can gather for longer.
Sorry to dig this up so late, there is a usual misconception when reasoning with power (which you might suffer from) :
What matters is not power, what matter is what we measure. We are measuring electric field in volts whose square is proportional to the power. Going 10x as far, mean measuring a voltage 100x smaller. Even with just the dynamic range if your analog digital converter is reading you 8-bits values (values going to 256) with a 100x voltage reduction you get a smaller blip going from 0 to 2.
If you have SNR of 6dB over some integration interval which gave you acceptable results, and you have 40dB of additional path loss, you're now -34dB, and you need to make that up. There's no "taking half" because we're talking about only a 100x difference in receive voltage.
Put another way, we already "took half"-- our SNR of 6dB (quadruple the power) was already only double the voltage. Our metrics for SNR already take what you're talking about into account.
Dynamic range can be a factor, but you generally have some kind of automatic gain control that increases effective dynamic range (so that in absence of signal, you have noise much bigger than 1LSB showing up at the converter), and the conversion dynamic range "width" of the converter only matters when there are other in-band transmitters around you need to reject (because they limit how far you can turn up the initial gain before saturating the converter).
Note also, not that it relates to what we said at all-- you can detect signals much smaller than 1LSB because of dither.
Yea, that's not new with the 60Ghz, it's already possible with wi-fi signals, even the pose can be estimated:
And here I'm not even considering if this solid state radar chip cross reference its data with Wifi radio signals or the dot projector, boosting its ML recognition capabilities to be able to "blob" (and possibly identify) moving things as far as 50m.
The main reason it was chosen was to make it smaller to have everything integrated on a chip, and have an antenna array.
Walls are typically static, which means they won't appear in the hardware amplified Doppler-shift signal (below 100Hz) which detect moving objects. (Radar works by amplifying the low frequency beating that occurs when adding two signal of high but close frequency 60GHz and 60Ghz+50Hz, low pass filtering then amplifying.)
Typically the limits about power are dictated by how much attenuation (distance + obstacles) in dB *4 because inverse 4th power law, but as long as you are above the background noise you can amplify.
Also here it is not trying to form a focused image, it is just gathering some signal to run some pattern detection algorithm.
Even if there is a lot of noise we can integrate it with software fft over a long period of time, because we are trying to look for breathing movements (0.3Hz) which are even slower.
It will kill your SnR entirely, though. There will realistically be nothing to amplify.
From the "Technology" page linked at the top.
Personally, I don't see how it is meaningfully any worse for privacy than the always on microphone in the google assistant. And I look forward to what it will enable for VR and AR tech!
"Never sending raw sensor data to Google" is an efficiency claim that means little with regard to protecting privacy.
It doesn't seem like the kind of thing a company would try to underhandedly sneak in after explicitly saying the sensor runs on-device and "never sends sensor data to Google servers".
* Trivial in the sense of someone who actually knows what they're doing -- of which we only really need one person in order to "leak" what's actually happening.
I don't think anyone thinks that they are, or will be sending all the sensor data to the servers. But they don't need to, they only need to send back what is determined due to that data.
Specifically consider what is most likely for an ad company, a few differentiated markers to help with targeting, that can easily be packed into a few bits. And can be trivially stored to be sent later on with some other packet that has a valid reason to be sent. (Appstore update check, anyone?) Accounting for every bit that leaves the phone sounds like a nigh impossible task.
I'm not saying this is definitely happening, but I don't think it's practical to rule out the possibility, with any amount of packet sniffing.
Because I'm pretty sure there is some random "user experience checkbox" checked by default that somehow means some of the data is sent to Google so they can "improve their products and services" and what not, but don't worry because it is probably "anonymized" and will only be seen by "humans" or "AIs" depending on what feels less bad for the general public once discovered.
Would you by any chance be interested in buying a bridge in Brooklyn?
About 60Ghz technology : https://www.embedded.com/why-60ghz-mmwave-is-moving-into-the...
I've played with some 6Ghz 2$ microwave radar sensor and Software Defined Radios though I have not yet used 60Ghz.
Electromagnetic Waves are kind of magic when used in not conventional ways. I remember seeing the 2015 disney EM-sense video https://www.youtube.com/watch?v=fpKDNle6ia4 which is a non-radar way to listen to the environment.
I don't see why this wouldn't be an even better way to do the same thing.
... and of course it exists. Only $90: https://silent-pocket.com/products/faraday-cage-phone-sleeve
That will be, um, interesting.
>* Headed South is an experience that introduces Pixel’s new touchless interface to users in a playful and engaging way. Through your journey, learn, practice and master the use of Motion Sense gestures to gather a flock of birds and fly South to escape the storm.
Radar uses electromagnetic waves (like Wifi but higher frequency) so it can go through walls, and even typical range for gesture recognition is less than meter, It probably can go at least 10 times as far by boosting the gain of the amplifier, it is not constrained like a theremin would be because it is already working in the far field region of the antenna.
Because it work at such high frequency (but not so high that it can still go through walls), it has many very small antennas arrays, and sense sub millimeter movements even from far away. It also has beam-forming capabilities, meaning it can focus the direction in which to sense. Because it is radar, things which moves are of interests and are filtered easily from the background.
Typically this piece of technology already can or will soon be able to : sense how many humans are around it, where they are, how fast they breath, how fast the heart is pulsating, who they are by computing some heart-base ID.
It is low-power and always-on, 360° with focus-able attention. It is cheap because it can be made on a chip. (Edit: fixing typos)