NSA Releases Historical Documents on TEMPEST 121
sgunhouse writes to mention Wired's Threat Level has a piece on a recently-declassified document detailing the history of TEMPEST. "It was 1943, and an engineer with Bell Telephone was working on one of the U.S. government's most sensitive and important pieces of wartime machinery, a Bell Telephone model 131-B2. It was a top secret encrypted teletype terminal used by the Army and Navy to transmit wartime communications that could defy German and Japanese cryptanalysis. Then he noticed something odd. Far across the lab, a freestanding oscilloscope had developed a habit of spiking every time the teletype encrypted a letter. Upon closer inspection, the spikes could actually be translated into the plain message the machine was processing. Though he likely didn't know it at the time, the engineer had just discovered that all information processing machines send their secrets into the electromagnetic ether."
This is gonna be interesting. (Score:5, Interesting)
Maybe the silliest consequence? (Score:3, Interesting)
Saw this a few days ago (Score:4, Interesting)
http://cryptome.org/ [cryptome.org]
nsa-spectrum.zip + Zipped NSA Cryptologic Spectrum Articles 1969-81 April 24, 2008 (31MB)
nsa-tempest.pdf + TEMPEST: A Signal Problem (NSA History) April 24, 2008
No direct link to save JY's bandwidth.
I love the simple solution
"Instead of buying this monster, the Signal Corps resorted to the only other solution they could think of. They went out and warned commanders of the problem, advised them to control a zone about 100 feet in diameter around their communications center to prevent covert interception, and let it go at that."
I am trying to get some time to get into the Spectrum articles.
Re:Maybe the silliest consequence? (Score:4, Interesting)
Unless you are using a fiber optic cable with a transparent sheath [flickr.com], there shouldn't be any kind of detectable emissions from a fiber optic cable, especially not EMF, since there shouldn't be any moving electric current, right?
The line might heat up very slightly from the signal losses, but that wouldn't be rapid enough to reveal anything useful about the signal, especially if manchester encoding [wikipedia.org] is used, where the light would be on 50% of the time.
Something you can try at home... (Score:5, Interesting)
You're also going to need a portable black-and-white TV, a decent aerial, and maybe an aerial booster.
Testing is simple - put a recognisable image up on the screen. This can be the startup screen of an application, a directory listing, even an ASCII-art goatse if you're so inclined. The key is is *must* be a monochrome screen with pixels that are on or off - it won't work with greyscale. There's a subtle side-effect of this, which I'll come to in a moment.
Plug the aerial into the black and white TV. If you're more than a few feet away from the target computer, you're going to need the aerial. The signal you're trying to pick up is *tiny*. Tune the TV until you see what looks like a garbled version of the computer screen - an analogue tuner is best for this. The picture will be extremely weak and noisy, and it will also not be synchronised correctly. Now adjust the horizontal and vertical hold on the TV until you get a stable picture. You should at least be able to make out roughly what's on the screen.
To take it further, you need to break into the TV and add an AM radio. This detects the scanning coils in the monitor, and allows you to generate a sync pulse to lock the TV to the computer. You need to position the TV and AM radio very carefully so the radio isn't picking up the TV scan coils. This is the difficult bit, and in fact I've never got this part to work. I've got readable text off the computer screen before, from about 30 feet. I'd call that working.
Back to the greyscale thing briefly - antialiased fonts use grey pixels on either side of the black or white pixels to "blur" the edges and make the fonts look smoother. This has the effect of lowering the rise time of the signal, and thus not throwing as many harmonics out. Think about it - a switch from a black background to a white pixel is basically a squarewave, but if you step through a couple of shades of grey there's a much lower amplitude change and so the harmonics will be correspondingly quieter. So, anti-aliased fonts prevent Tempest-style attacks, and in fact about 15 years ago you could get "Tempest Fonts" that were basically very fuzzy antialiased fonts.
The other thing is that LCDs don't emit RF harmonics to nearly such an extent. The days of Tempest and Van Eck phreaking are pretty much gone.
TVs picking up Laptop Signals (Score:4, Interesting)
Since I'd done work with TEMPEST in the 1980s and was hanging out with a bunch of crypto people, and since the open-source discussions were mostly people saying "Laptops should protect you just fine since they're LCD", I obviously had to speculate about how this could be happening. My guess is that it wasn't the LCD itself that was radiating, but instead was the VGA jack on the back for plugging into a desktop monitor. Most laptops still have those today, and while many people use LCDs rather than CRTs as desktop monitors, they're still connecting by VGA signals using not-particularly-shielded cables, so there should still be plenty of signal around to listen for.
Obviously today's video signals are a lot higher frequency, so you'd need to use some actual computer equipment rather than squinting at a television. I don't know if the digital signal formats are easier or harder to intercept successfully than the VGA analog ones; maybe that'll help.
Re:I guess that is why some of the offices.... (Score:3, Interesting)