Slashdot Mirror
Matchstick-Sized Sensor Can Record Your Private Chats Outdoors
wabrandsma sends this story from New Scientist:
"A sensor previously used for military operations can now be tuned to secretly locate and record any single conversation on a busy street. [A] Dutch acoustics firm, Microflown Technologies, has developed a matchstick-sized sensor that can pinpoint and record a target's conversations from a distance. Known as an acoustic vector sensor, Microflown's sensor measures the movement of air, disturbed by sound waves, to almost instantly locate where a sound originated. It can then identify the noise and, if required, transmit it live to waiting ears. Security technologist Bruce Schneier says this new capability is unwelcome – particularly given the recent claims about the NSA's success at tapping into our private lives. 'It's not just this one technology that's the problem,' Schneier says. 'It's the mic plus the drones, plus the signal processing, plus voice recognition.'"
While the sensor measures the vector direction of sound, it measures the *SUM* of all sounds impinging on that point. With a single point sensor, you can't separate the direction of a particular one.
Now, a reasonably small array of these sensors (maybe as small as 30cm across) might be able to accurately localize the source, in the same way that your brain can tell what direction a sound is coming from with just two ears.
Of course, it will also be confused by multipath and reflections, although if adequately characterized, those could be used for localization as well, since the reflected paths have different spectral properties.
There are LOTS of sensors that are tiny and easily deployed these days, using all sorts of sensing modalities. And processor horsepower and tiny comm to get the signals to the processor are easier to get too.
Governments have always had the technical means to be invasive, they are restrained by common decency and the law of the land. I actually would have more concerns about private industry: they're not subject to most privacy laws (at least in the U.S.): The wiretap laws, for instance, refer only to comms over a system. If you want to set up a big acoustic array at the beach and record everyone's conversations, process it, and sell the product to whoever, that's perfectly legal. Or set that array up out on the pubic street in front of your house, or the thermal imaging camera, or the mmWave wall penetrating radar or whatever.
So, it's those laws and common decency is where the efforts should concentrate.
Governments have always had the technical means to be invasive, they are restrained by common decency and the law of the land.
What governments would those be which are restrained by the law, let alone common decency?
"They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety."
No, having the lack of privacy go both ways isn't as good as having privacy. A system where nobody can keep their actions private is a system governed by mob rule, nobody wants to engage in socially unacceptable behavior because they're instantly shunned and those who fail to participate in the shunning are also shunned for condoning it. Nobody will speak out unpopular opinions even if they feel it ought to be said, because those who don't like the message will go after the messenger. All social circles become totally transparent and people will self-censor their associations to avoid social stigma. It's freedom of the "you have freedom of speech and can say what you want, but we'll shoot you afterwards" variety.
Not to mention, it won't work. The powers to be will always find some reason why their conversations must be protected in the name of national security - after you've given up yours in the name of national security, of course. And if you don't like it you've got something to hide and is probably one of the bogeymen we're trying to catch. They can clam up any time they feel like it, while you'll stay stripped bare. Only the truly naive wants to head us in that direction, because <Admiral Ackbar>It's a trap!</Admiral Ackbar> and a pretty obvious one at that.
Live today, because you never know what tomorrow brings
That's true for a single snapshot of sound in any given instant. But if you collecting acoustic data over a period of time, transient sounds (noise) average out, and the loud peak (gunshot) or cyclical nature (engine) of certain sounds of interest should ease their extraction, with sufficient processing.
That was my impression too (one of my grad school courses was on acoustics and sonar design). That's what they do in submarines - make a great big phased array microphone. But if you follow the link in TFA to the company's site, they have a PDF which gives a bit more info on how their sensors work:
So whereas a phased array sensor works by comparing the arrival times of a wavefront at different locations to determine the direction the wavefront is traveling, it sounds like their gizmo is measuring in 3 dimensions the actual movement of air molecules caused by the sound wave, and deriving the wavefront travel direction from that.
One difference I can think of between electromagnetic radiation and sound waves is that the former are transverse waves and the latter are longitudinal waves. Does that make much of a difference for these purposes?
It's been explained already by Solandri but I'll give it a try. The sensor measures the displacement of the air, not the pressure. The GP post argues that it measures the vector sum of the displacement caused by all sounds, and this is correct. I imagine a small light ball that's magnetically suspended and being pushed around by the air, and the movement of the ball is recorded (this is probably not how it's done, I imagine it would have a highly non-linear frequency response). If there are two sounds from different directions with different frequencies, you could easily tell them apart -- e.g. the ball is moving up and down fast, and left and right slowly. If the sources have the same frequency, or it's just some broadband noise, you can't tell them apart. You could probably do it easily with two vector sensors and relative timing, but the whole point was that you can do it with one sensor in plausible conditions.
As for electromagnetism / sound, you are right that the transverse/longitudinal distinction makes a difference. The air displacement is analogous to the electric field. An RF antenna sort of measures the projection of the electric field onto some given axis, and an omnidirectional antenna measures approximately the absolute value of the electric field. (A mic may measure pressure, not displacement, but these are alternative variables for discribing the same wave). An analogous RF detector to the sound vector sensor would be three small linear antennae pointing in linearly independent directions, measuring the x,y and z components of the electric field as functions of time. The transverse / longitudinal distinction comes in here: the EM waves have an additional degree of freedom, namely polarisation. The E field can point in any direction perpendicular to the direction of propagation. Given the polarisation vector you can thus only constrain the vector to the source to a plane.
I just got back from a pleasure trip. I took my mother-in-law to the airport.