Through-Wall Human Pose Estimation Using Radio Signals

I believe you can achieve the same result in your home by adding a sheet metal roof, metal cabinets, and refrigerators along every wall.

FWIW, if they are intentionally (or unintentionally) blocking your cell signal with an electronic signal (And since you mentioned that they didn't remodel, this is the only way they could feasibly do it), they're breaking a very serious FCC law - https://www.fcc.gov/general/jammer-enforcement.

I know that the FCC is kind of regarded as a joke lately, but this is a law they take pretty seriously as jammers can interfere with E911 calls. It would just take one emergency that 911 couldn't respond to for the FCC to realize what was going on, and no large store is going to take those odds. It's likely something else.

That said, if you really believe that they are blocking cell signals, you should probably file a tipoff, as they could honestly be putting lives at risk (go to https://consumercomplaints.fcc.gov/hc/en-us and click "phone"). As far as I know there's no penalty if you are wrong (as long as you honestly believe there might be an active jammer in the area and you aren't abusing their system somehow), worst case you'll get a call confirming they don't use one.

If you could publish evidence of this phenomenon, the public would be very interested.

Having discussed this with a couple of friends off of here, I should add some evidence of the tracking[0]. I agree with lgats sibling comment that this is likely just a function of the construction of the building and not outright illegal jamming.

[0] https://www.walkbase.com/blog/wi-fi-analytics-in-retail-stor...

Anecdotally, I've had the same experience. Been going to Home Depot for a number of years, routinely using the app to locate the aisle I want in the store. Starting less than a year ago my signal becomes almost unusably weak as soon as I walk through the doors. No construction on the store, same store as has been there for a decade, but something changed. Might be a coincidence.

I've noticed this, but my thought is that it is just the natural faraday cage created by having a cement building filled with metal shelves. Combine that with towers transitioning to use LTE over 3G and you get reduced coverage indoors in large buildings.

classic hackernews

In terms of passive signal blocking, I wonder if the chickenwire mesh that holds exterior stucco in place could serve as an effective Faraday cage? Or aluminum siding? I'm sure the metal roof and walls of a modular storage building would.

Most businesses use metal studs to support interior walls, though I doubt their ~2 foot spacing would block wifi or cellular RF.

In terms of active cellular jamming, that should be pretty easy to detect -- by standing in an outside doorway where you'd be exposed to inside AND outside signals, and then checking if your cellular bars fade out. That should indicate that you've been exposed to an internal jamming signal, since the external signal should still remain.

It's hard to imagine that the advantage to a corporation of isolating the customer's phone would be worth the risk of FCC prosecution.

Could it be that they had repeater antennas before, but after the LTE upgrades, they either declined to renew the repeater leases in favor of wifi+tracking or perhaps the carriers didn't want to pay for hosting repeaters?

I haven't noticed it myself. Cell phone still works inside my local home depot.

Back in college the engineering building used metal plates as the outside facade, combined with concrete construction. It had a side effect of getting rid of any cell phone reception as a result, unless you were beside a window. The building was built before 2G cell phone service was a thing, so it wasn't an intentional thing.

Go to hardware store (Home Depot would be ok but watch for poor cell reception). Buy 100 rolls of chicken wire and several gross of outdoor cable clips. Tack up all over outside surface of your house.

Steel roof trusses, steel roof material, and racks and racks of steel shelving don't exactly make for an ideal radio signal environment. They aren't blocking signal, you are standing in an un-intentional Faraday cage. It likely has more to do with who your carrier is and what frequency your phone is connecting at. And FWIW, you don't have to be connected to a store's WiFi for them to track you - https://www.fastsensor.us/en/ is only just one of the many companies who don't require your cooperation to track you in-store.

It's somewhat similar, but AT&T (and Apple) have forced my iPhone to always join the "attwifi" public wireless networks (present in Home Depot, airports, etc.).

I can't make my iPhone NOT auto-join it, no matter how many steps, guides, hop-on-one-foot voodoo that I've tried.

The wireless network is terrible at everything except signal strength.

A common cause for this is foil backed insulation. A number of companies install this in the ceiling spaces and outside walls to reduce the energy costs of keeping the store at a comfortable level.

You could wrap your house in chicken wire and ground it. I think that's about as cheap and easy as it can get. Might make the HOA mad, though.

"Available for me". Are you assuming we know who you are? What's available for you might be very out of reach to others. Farady cages are not some wallpaper you put on your house.

I always assumed stores were blocking cell signal (or certainly not doing anything to improve it, like installing LTE repeaters), to prevent you from going on Amazon and finding cheaper alternatives.

This seems nearly identical issue; why would you suspect the vote would differ?

So this MIT method would be OK, since it's not thermal?

FWIW, I don't think the cops were using thermal imaging to detect humans within a house - they were using it to detect abnormally high heat - from high temperature lamps used to grow pot. Originally they'd talk to the power company to see who is drawing irregularly high levels of power, but the growers started using diesel power generators to power their grow lights.

Sometimes I wonder is this could be mitigated by literally covering your walls in aluminum foil. Maybe those "tinfoil hats" are on to something.

"Attacks only get better".

Meh, that seems a little sensationalist. Any significant metal would likely throw this in a loop. You could even buy paint with graphite in it to block weak RF signals outright. IIRC most military bases and data centers protect against this sort of leakage.

The skeletons appear to have very long legs, but no torsos.

I wonder if it was made intentionally that way to make them look less serious.

Guess you shouldn't read about what synthetic aperture radar systems have been able to do for decades... you can actually make a ghetto radar capable of capturing the data used for this demo with about $350 (in 2012) in materials. But don't worry too much. Their limitations section states that "the operating distance of a radio is dependent on its transmission power" so their system only works reliably up to 40 feet, which is true, but the dominating factor is simply distance, because any signal you get back will be attenuated proportional to 1/r^4. Pumping up the transmitted power to compensate takes a lot of energy that quickly gets you into needing operating licenses. Not to mention moving to using different frequencies that can give higher resolution of whatever you're imaging...

No, this research documents information you are already leaking, it isn't taking anything away.

They already have your up-to-the-second location and all your private data. All this adds is what pose you're currently making with your body.

Speaking of, I was halfway expecting to see a couple stick figures banging each other behind a wall somewhere in that video. I really think that would have been the frosting on top; they should consider this for the next video when it's time for another round of funding.

I hope you're not doing anything illegal, citizen.

Maybe SWAT can use this to validate that it's actually the right house when serving a no-knock warrant. I'm cynical enough to feel like there's no way this is good for any of their targets, though.

Edit: Kolpa makes a really good point! This seems like a really good counter to SWATting, where the claim is that someone is holding hostages or something -- this could pretty effectively demonstrate that to be false if they see one person in an apartment sitting at a computer.

but thermal imaging is already available without AI or WiFi..

Was just thinking that this reminds me of the alien weapon in Perfect Dark: https://youtu.be/8DlVQ6OjRcA?t=6s

And this is why you build a shipping container house -- free Faraday cage, built right in. Not mention that it's not hard to convert shipping containers to meet California earthquake code, or so I've read.

So Dark Knight style imagery, all from 1 truck?

What's the range?

Potentially not that simple, as there are fundamental limits on the information capacity of a MIMO system, based on the surface area of the volume occupied by the antenna array (in wavelengths squared). The capacity limits apply whether you use the array for communications or radar, as both are governed by Information Theory.

Not to say that it can't be done, but the maths exists to definitively prove whether it can be, so no need for speculation and tinfoil. My gut feeling is that you could scan a small volume from afar, but not a whole block. No, I haven't done the maths.

Smelly air is "coming", so prepare for it by eating lots of onions? Nah.

> I've heard quite a lot of people that talk about post-privacy, and they talk about it in terms of feeling like, you know, it's too late, we're done for, there's just no possibility for privacy left anymore and we just have to get used to it. And this is a pretty fascinating thing, because it seems to me that you never hear a feminist say that we're post-consent because there is rape. And why is that? The reason is that it's bullshit.

> We can't have a post-privacy world until we're post-privilege. So when we cave in our autonomy, then we can sort of say, "well, okay, we don't need privacy anymore, in fact we don't have privacy anymore, and I'm okay with that." Realistically though people are not comfortable with that. Because, if you only look at it from a position of privilege, like, say, white man on a stage, then yeah, maybe post-privacy works out okay for those people. But if you have ever not been, or if you are currently not, a white man with a passport from one of the five good nations in the world, it might not really work out well for you, and in fact it might be designed specifically such that it will continue to not work out well for you, because the structures themselves produce these inequalities.

> So when you hear someone talk about post-privacy, I think it's really important to engage them about their own privilege in the system and what it is they are actually arguing for.

-- Jacob Appelbaum

I'm scared. What is a post-privacy world?

Describing the exact same inevitable phenomena as "The Age of Transparency" is probably going to elicit a better response from most people. "Post Privacy" is negatively charged and makes it sound like more of a net loss as opposed to progress.

It’s a new paper from the same group.

In 2015, the tracker person was just shown as an amorphous blob on the screen. In the recent paper, they display a stick figure modelling the movements of the person’s limbs.

They’re using custom hardware to make this work.

This isn’t always clear from the press release. For example, they mention that the technology works at “WiFi frequencies.” That’s true, but they’re sending a radar signal using two huge antennas, not a Wi-Fi router.

May I ask what's FMCW? Of course I can straightaway google but I would like to hear an intuitive explaination.

They are quite different. Wisee uses a single reciever and uses doppler shift to identify the gestures whereas this "through wall" system uses an array of transmitters and receivers that identify reflections from a position in space.

In layman's terms this system is more like a conventional camera but working in the radio spectrum whereas Wisee is like a microphone that can pick up changes in frequencies around it.

Duh.

You clearly didn't read the manual.

I know you're joking, but that's an interesting idea... I wonder how would the system "see" your head then? Unless it is trained on people wearing tinfoil hats (and aluminium suits) it might produce interesting results.

Does it matter if our WiFi points are always on?

maybe so... but only external walls right and even then dont you want wifi in the back yard?

>I'd be shocked if there wasn't already an existing capability to detect bad guys through walls in a CQB environment.

Existing and exist in a state you can deploy are two totally different things that often happen many years apart. Usually something exists as an impractical wish list item for a long time before some other development pushes it into the realm of being practical to use.

May I ask what's CQB?

Yeah, I wonder what they are going to use this technology for. "Helping earthquake victims" probably

I bet some RF absorbing material or something that reflects RF at weird angles would probably break it. Or an active RF transmitter.

anything metal in the way. I'm sure the image would be a lot less clear if the wall had metal studs instead of wood.

Make all walls a Faraday cage / concrete. If you can’t get cell or radio reception inside, you probably aren’t leaking anything either.

Build an electronic warfare device that makes it look like hundreds of people are inside, all pointing guns outwards.

Carry something metallic, so the neural network no longer classifies your reflection as human?

SWAT team dealing with a hostage situation in a closed room

A personal home security system could make better use of this instead of motion sensors in each room.

It can also be used to monitor a baby's respiration

As mentioned elsewhere, detecting earthquake victims maybe, a terrorist or hostage situation where the targets definitely aren't giving consent, military situations and covert surveillance for law enforcement and counter espionage all spring to mind as legitimate and useful applications.

Depends on the training data and the room config. The system sees a difference in an empty vs occupied room, movement doesn’t seem to be strictly necessary. The heat map video looks more like radio spectrum signatures, which might even be working on still frames - which would imply that movement isn’t necessary at all. It would help the AI, though. Easier to classify moving bodies, the speed and gait is a signal.

A good place to start would be the paper's references.

I built and extended this for a college project years ago: https://ocw.mit.edu/resources/res-ll-003-build-a-small-radar...

It's not a bad place to start. And cheap if you want to build your own too (some third party even sells the "cantennas" so you don't have to fiddle with tuning that part (which is kind of infeasible without a network analyzer anyway)).

Why can't it be both? Science and technology advance in often unexpected directions and for new purposes all the time. Satellites in orbit came from ICBM research, GPS from military location tracking, the tiny electronics in your phone are the grandchildren of at least one miniature spy device.

No we can't use this radio signals to see tumors, because science is not magic and wishful thinking will not solve the technical problems. Let's try to name a few:

The radio signals are absorbed by wet salty meatbags, so you can find them using radio signals, but you can't see thru them. If you raise the power to compensate the absorption, you will cook the meatbags, like a microwave oven.

If you were lucky to get a signal at the other end, the radio signal can distinguish between meat-bone-air-water, but not between tissues like meat and tumors that have a somewhat similar composition. A tomography use contrast to try to distinguish them. A magnetic resonance is better, but it use a huge magnet (and in some cases it use contrast too).

And also, the radio frequency that they are using has a wavelength of an inch approximately. So roughly it can resolve things that are an inch long, so it must be a big tumor to be visible. Perhaps you can try with a higher frequency that has a smaller wavelength.

And if I understand correctly, this device needs a lot of calibration, so you need version of the part of the body without the tumor and a version with tumors of different sizes and positions to calibrate the device and then try to use it in the real subject.

Devil's advocate: "let's fund PhDs with respectable salaries instead of green cards."

Why do you believe any of those names are not American?